JP2841453B2 - Image transmitting apparatus, image transmitting method, image decoding apparatus, and image receiving method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image transmission apparatus and an image transmission method for hierarchically encoding and transmitting a still image signal.

[Conventional technology]

Hereinafter, a conventional still image transmission device will be described with reference to FIG. (1) is a microcomputer, CP
U (central processing unit) (2), ROM (3) and RAM (4). (5) is a bus (consisting of a data bus, an address bus, a control bus, etc.) of the CPU (2). The microcomputer (1) controls each unit of the still image transmission device. (13) is a transmission line, which can be wireless or wired. In the case of a wired transmission line, ISDN (Integrated Services Digital Network), high-speed digital line, analog telephone line, DDX (Digital Data Exchange network (this includes DDXC,
DDXP) and dedicated lines.

(12) a communication processing circuit connected between the transmission line (13) and the bus (5) for performing signal processing according to the protocol and transmission speed of the still image signal of the transmission line (13); In the interface, the communication processing means transmission processing and reception processing such as encoding and modulation for transmission and decoding and demodulation for reception, respectively.

(6) is a frame memory (video memory) whose digital video signal input / output terminal and control signal input terminal are connected to the bus (5), and whose digital video signal input terminal is
The digital video signal output terminal is connected to the input terminal of the D / A converter (7) and the other control signal input terminal is connected to the output terminal of the A / D converter (10). Connected to output terminal of circuit (11). The video memory (6) includes a microcomputer (1)
Controls writing and reading.
The video memory (6) includes a horizontal and vertical address counter, a memory controller, and the like.

(9) is an input terminal for an analog video signal. A video signal (video signal from a video camera, a VTR, etc.) from this input terminal (9) is supplied to an A / D converter (10) and converted into a digital video signal. After the conversion, it is supplied to the video memory (6) and written.

(14) is a communication memory, that is, a buffer memory for transmission and reception. The writing and reading of the communication memory (14) are also controlled by the microcomputer (1). The communication memory (14) includes a horizontal and vertical address counter, a memory controller, and the like.

(15) is a digital signal processing circuit (DSP) capable of high-speed signal processing, and has an external RAM (16) and an internal RAM (17)
Which is also controlled by the microcomputer (1).

The analog video signal supplied to the input terminal (9) is
After being supplied to a D converter (10) and converted into a digital video signal, it is supplied to a video memory (6) and written as a digital still image signal. The still image signal stored in the video memory (6) is block-coded by the microcomputer (1) and the digital signal processing circuit (15) for each block signal, that is, orthogonal coding such as discrete cosine transform is performed. After the conversion and the subsequent compression coding by variable length coding such as Huffman coding, the data is written to and read from the communication memory (14).
After being supplied to the communication processing circuit and the interface (12) and subjected to transmission processing, it is transmitted to another still image transmission device via the transmission line (13).

The transmission signal supplied from another still image transmission device to the communication processing circuit and the interface (12) through the transmission line (13) is received and processed here, and the obtained orthogonal transform and compression encoding are performed. The digital video signal is written into the communication memory (14), where it is block-decoded by the microcomputer (1) and the digital signal processing circuit (15), that is, the orthogonal inverse transform and the subsequent decompression decoding are performed. After that, each obtained block signal is written to the video memory (6) to form a still image signal. Then, the digital video signal read from the video memory (6) is supplied to a D / A converter (7) and converted into an analog video signal, and then a monitor receiver (8) having a cathode ray tube is provided. And is projected as a still image on the surface of the cathode ray tube.

Next, referring to FIG. 7, IEEE TRANSACTIONS ON COM
MUNICATION (IEE Transaction on Communication), VOL.COM-32, NO.3, MA
A case will be described in which the encoding by the adaptive discrete cosine transform (ADCT) disclosed in P.225 to P.232 of RCH 1984 is applied to the still image transmission apparatus shown in FIG.

As shown in FIG. 7A, a digital video signal for one frame (here, the video signal is a monochrome video signal for simplicity of description) stored in a video memory (frame memory) (6), That is, 768 × 480 8-bit pixel signals forming a matrix of 480 rows and 768 columns are converted by the microcomputer (1) as shown in FIG.
The digital signal processing circuit (15) divides each block signal into 8 × 8 block signals composed of 8 × 8 adjacent pixel signals (indicated by small rectangles) forming a matrix of 8 rows and 8 columns, and reads out each block signal. Write to external RAM (16).

Then, each block signal written in the external RAM (16) is converted by the digital signal processing circuit (15).
Performs two-dimensional discrete cosine transform (two-dimensional DCT). A generalized version of this is shown below by mathematical expressions.

j, k (where j, k is j, k = 0,1,2,3, ..., N-
The two-dimensional discrete cosine transform F (u, v) of the sequence f (j, k) of 1) is expressed by the following equation.

Where u, v = 0,1,2, ..., N-1 The 8 × 8 pixel signals forming the 8-row × 8-column matrix stored in the external RAM (16) are obtained by performing a two-dimensional discrete cosine transform on the eight rows shown in FIG. 7C. 8 × 8 coefficient signals forming an eight-column matrix (for example, 12
(Rounded to bits) (indicated by a small rectangle) is a DC coefficient (DC) (average value of 8 × 8 pixel signals) in the upper left corner
As the signal arrives and moves away from it in the horizontal and vertical directions, a low to high frequency coefficient signal will be distributed.

And, by the microcomputer (1),
The coefficients of the 8 × 8 coefficient signals forming the matrix of 8 rows and 8 columns stored in the RAM (16) correspond to the quantization divisors forming the quantization matrix of 8 rows and 8 columns selected in advance. Quantization is performed by dividing the quotient, and the quotient is rounded so as to be represented by 8 bits. FIG. 7D shows an 8 × 8 quantized coefficient signal (shown by a small rectangle) constituting a matrix of 8 rows and 8 columns obtained by this, and the quantized DC coefficient at the upper left corner thereof is shown. As the distance from the signal (DC) increases, coefficient signals having a coefficient of 0 frequently appear, and in many cases, reach 2/3 of all quantized coefficient signals.

Then, 8 × 8 quantized coefficient signals (shown by small rectangles) shown in FIG.
7, the zigzag scan is performed as shown in FIG. 7E except for the DC coefficient signal at the upper left corner (or may be included), and then compression encoding (Huffman encoding) is performed, and the communication memory (14) shown in FIG. ). The DC coefficient signal of the quantized coefficient signal is directly compressed and encoded (Huffman encoded) and then written into the communication memory (14).

The IADCT of the digital video signal encoded by ADCT
Although a detailed description of (inverse adaptive / discrete cosine transform) will be omitted, in brief, a digital video signal encoded by the ADCD stored in the communication memory (14) is converted to a microcomputer (1). )
Decompression decoding (inverse Huffman encoding), inverse zigzag scanning, inverse quantization, and two-dimensional inverse cosine transform of the signal by a digital signal processing circuit (15) to obtain a block signal. The block signal is written into the video memory (6) by the microcomputer (1), and this is repeated to form a still image signal.

In the above-described conventional still image transmission apparatus, a coarse still image signal is created from the original still image signal to be transmitted, and transmitted to the other still image transmission apparatus, and the transmitted still image signal is interpolated. Processing, and also generates a corrected difference signal for the interpolated still image signal, and transmits the corrected difference signal to the still image transmission device of the other party once or divided into a plurality of times. If the modified difference signal is added to the still image signal of the coarse still image to correct the signal, the still image signal is coarser than the case where the still image signal is compressed and encoded as it is and transmitted. And the coarse still image is then gradually modified to a finer still image.

Therefore, next, a description will be given of a conventional still image transmission apparatus in which hierarchical encoding and layer decoding are added to compression encoding and decompression decoding in the conventional still image transmission apparatus, respectively.
Since the overall configuration of the still picture transmission apparatus is substantially the same as that of the above-described still picture transmission apparatus of FIG. 6, the still picture transmission apparatus of FIG. 6 performs hierarchical compression encoding and hierarchical decompression decoding. The case of application will be described with reference to FIG. 8 and FIG. In the following description, the case of the luminance signal will be described, and the description of the case of the red and blue difference signals will be omitted.

FIG. 8 is an explanatory diagram of the conventional hierarchical encoding and hierarchical decoding, and FIG. 9 is an explanatory diagram for assisting the explanation.

Each memory in FIG. 8 is one of a plurality of storage areas of a buffer memory provided separately from each memory shown in FIG. 6, and the buffer memory is a bus (5) shown in FIG.
Shall be connected to

In the encoding system (hierarchical compression encoding system) of FIG. 8, a digital luminance signal for one frame stored in the video memory (frame memory) (6) of FIG. 6 under the control of the microcomputer (1). (Still image signal) (This is a matrix of 480 rows and 768 columns, each of which is 8 bits 768 × 4
FIG. 9A shows a set of rectangles indicating 16 × 12 pixels constituting a matrix of 12 rows and 16 columns corresponding to the pixel signals. 9B and onward, so as to facilitate understanding of the description of the hierarchical compression encoding and the hierarchical expansion decoding in FIG. 8), and read them out from the input terminal (20) of the encoding system. And low-pass filtering and 1/2 decimation processing (meaning two-dimensional low-pass filtering processing and subsequent 1/2 decimation processing in horizontal and vertical directions (subsampling), and so on) (2
After 1), the data is reduced and written to the memory (22). One frame of digital luminance signal (consisting of 384 × 240 pixel signals forming a matrix of 240 rows and 384 columns of 8 bits each) written to the memory (22) and halved is read out. After performing low-pass filtering and 1/2 thinning-out processing (23), adaptively spaced cosine transform (ADCT) (24) is performed.

The above-described low-pass filtering / half-thinning process (21), writing and reading in the memory (22) and the low-pass filtering / half-thinning process (23) are basically equivalent to the video memory (frame) shown in FIG. 1) stored in the memory (6)
A digital luminance signal (still picture signal) for a frame (which is composed of 768 × 480 pixel signals forming a matrix of 480 rows and 768 columns of 8 bits each) (FIG. 9A) is read out. As shown in FIG. 9B, low-pass filtering and 1/4 decimation processing (meaning two-dimensional low-pass filtering processing and 1/4 decimation processing in each of horizontal and vertical directions, and the same applies hereinafter) (No. 9)
If this is reduced and written into the memory, it is thinned out to 1/4 consisting of 192 × 120 pixel signals of 8 bits each in 120 rows and 192 columns (FIG. 9C). Thus, a digital luminance signal (still image signal) for one frame is obtained. Then, the digital luminance signal (still image signal) for one frame which has been thinned out to 1/4 is subjected to adaptive discrete cosine transform (ADCT) (24).

The ADCT (24) is composed of 8 bits each having 192 rows of 120 rows and 192 columns.
A digital luminance signal (still image signal) for one frame, which is composed of × 120 pixel signals and thinned by / 4, is once written in the above-mentioned buffer memory, and is described with reference to FIG. Similarly, the microcomputer (1) divides the signal into block signals, and performs a two-dimensional discrete cosine transform on each block signal by a digital signal processing circuit (15). The microcomputer (1) performs quantization, zigzag scanning, and variable This means that long coding (Huffman coding) is performed.

In this way, the adaptive discrete cosine transform (ADCT) (24) outputted to the output terminal (25), the digital luminance signal (still image signal) for one frame, which is decimated by 1/4, It is transmitted to the still image transmission device on the side, and is subjected to decompression decoding and interpolation by the decoding system. This will be described below. A digital luminance signal (still image signal) for one frame that has been subjected to adaptive discrete cosine transform (ADCT) and decimated to 1/4 from an input terminal (39) is subjected to inverse adaptive discrete cosine transform (IADCT) (40 ) Is done. This IADCT (4
0) is subjected to decompression decoding (Huffman decoding), inverse zigzag scanning, inverse quantization, and two-dimensional inverse cosine transform by the microcomputer (1) in the same manner as described above, and each is returned to the original block signal. From these block signals, a digital luminance signal (still image signal) for one frame, which is thinned out to 1/4 consisting of 192 × 120 pixel signals (FIG. 9D) of 120 bits and 192 columns of 8 bits each. ) Is obtained and written to the buffer memory.

Each of the 8 bits stored in this buffer memory
A digital luminance signal (still picture signal) for one frame, which is composed of 192 × 120 pixel signals of 120 rows and 192 columns (FIG. 9D), is decimated by 1/4 and controlled by the microcomputer (1). , 2/1 interpolation processing (meaning 2/1 interpolation processing in each of the horizontal and vertical directions, the same applies hereinafter) (41), and then enlargement processing is performed and written into the memory (42). A digital luminance signal for one frame interpolated by 2/1 written in this memory (42) (each consisting of an 8-bit matrix of 240 rows and 384 columns)
(E of 384 × 240 pixel signals) (FIG. 9E) is read out, and 2/1 interpolation processing (43) is performed.

The above-mentioned 2/1 interpolation processing (41), writing and reading in the memory (42), and 2/1 interpolation processing (43) are, in short, the 8-bit data stored in the buffer memory.
A 1/4 thinned digital luminance signal (still image signal) (FIG. 9D) composed of 192.times.120 pixel signals forming a matrix of 120 rows and 192 columns is read out. 1 interpolation processing (means 2/1 interpolation processing in each of horizontal and vertical directions,
The same applies to the following.) If this is enlarged and written in the video memory (6) in FIG. 6, 768 × 480 pixel signals (8 × 480 × 768 pixels) constituting a matrix of 480 rows and 768 columns are obtained. A digital luminance signal (still picture signal) for one frame which has been subjected to the interpolation processing shown in FIG. 9F) is obtained.

If the interpolated digital luminance signal (still image signal) for one frame is reproduced by the monitor receiver (8) in FIG. 6, the still image can be seen promptly although it is a coarse still image. Can be.

Next, a first method for correcting the still image signal of the coarse still image is described.
The creation of the modified difference signal and its encoding and decoding will be described.

First, in the encoding system, the adaptive discrete cosine transform (ADCT) (24) output digital digital signal (still image signal) for one frame, which is output to the output terminal (25), is decimated to 1/4. , Inverse adaptive discrete cosine transform (IADCT) (26) [same as inverse adaptive discrete cosine transform (IADCT) (40) of the decoding system].
192 × 1 which constitutes the original 8-bit 120-by-192 matrix
A digital luminance signal (still image signal) for one frame (still image signal) (FIG. 9D), which is made up of 20 pixel signals and thinned by 1/4, is obtained and written into the buffer memory.

Each of the 8 bits stored in this buffer memory
One frame of digital luminance signal (still image signal) consisting of 192 × 120 pixel signals of 120 rows and 192 columns and thinned to 1/4
(FIG. 9D) performs 2/1 interpolation processing (27) [same as 2/1 interpolation processing (41) of the decoding system] under the control of the microcomputer (1), and then expands the memory ( Write to 28). One frame of digital luminance signal (8 bits each) interpolated by 2/1 written to this memory (28)
The 384 × 240 pixel signals forming a matrix of 240 rows and 384 columns (FIG. 9E) are the same as those stored in the memory (42) of the decoding system, and this is the memory (22). , A digital luminance signal for one frame which is thinned out by half (384 bits constituting a matrix of 240 rows and 384 columns of 8 bits each)
× 240 pixel signals), and are subjected to subtraction filtering (29) and interpolated by 2/1 interpolation processing (41) in the decoding system for one frame of digital luminance signals (240 rows of 8 bits each of 384 rows, 384 pixels). A first modified difference signal (consisting of 8 bits each) consisting of 384 × 240 pixel signals constituting a matrix of columns
(G of FIG. 9) comprising 384 × 240 difference pixel signals forming a matrix of 240 rows and 384 columns.

Then, the first modified difference signal is subjected to adaptive discrete cosine transform (ADCT) (30) for each block signal,
The modified differential signal subjected to the adaptive discrete cosine transform output from the output terminal (31) is transmitted to the still image transmission device on the other side.

In the decoding system, the adaptive differential cosine transform (ADCT) modified difference signal from the input terminal (45) is subjected to inverse adaptive discrete cosine transform (IADCT) (46) to obtain a first modified difference signal (FIG. 9). G), which is read out from the memory (42) and is a 2 / 1-interpolated digital luminance signal for one frame (384 × 240 pixels forming a matrix of 240 rows and 384 columns of 8 bits each) (FIG. 9E) and addition filtering (47), followed by 2/1 interpolation processing (48) and output to an output terminal (49). The output is enlarged and processed as shown in FIG. If written in the video memory (6), 768 × 480 pixel signals of 480 rows and 768 columns of 8 bits each (FIG. 9H)
Thus, a digital luminance signal (still image signal) (FIG. 9H) for one frame, which has been subjected to the correction processing, is obtained.

If the interpolated digital luminance signal (still image signal) for one frame is reproduced by the monitor receiver (8) in FIG. 6, the coarse still image described above will be corrected somewhat finely. According to this correction, not only the roughness of a still image by interpolation but also the roughness by ADCT and IADCT can be corrected.

Next, creation of a second modified difference signal for modifying the coarse still image signal, and encoding and decoding thereof will be described.

First, in the encoding system, the adaptive modified cosine transform (ADCT) (30) first modified difference signal output to the output terminal (31) is subjected to inverse adaptive discrete cosine transform (IADCT) (32) [decoding The same as the inverse adaptive discrete cosine transform (IADCT) of the system (46), thereby returning the original first modified difference signal to the 2/1 read from memory (28). The interpolated digital luminance signal for one frame (comprising 384 × 240 pixel signals forming a matrix of 240 rows and 384 columns of 8 bits each) (FIG. 9E) is added and filtered (33). .

The result of the addition filtering (33) is the 2/1 interpolation processing (3
4) After being enlarged, the data is enlarged and written to the memory (35), which is read out and read from the input terminal (20).
One frame of digital luminance signal (still image signal) stored in the video memory (frame memory) (6) in the figure.
(FIG. 9A) is subtracted and filtered (36) and interpolated by a 2/1 interpolation process (48) in a decoding system. The second modified difference signal (consisting of 768 × 480 pixel signals constituting a matrix of 480 rows and 764 columns of 8 bits each) for the constituting 768 × 480 pixel signals (FIG. 9I) Is obtained.

Then, the second modified difference signal is subjected to adaptive discrete cosine transform (ADCT) (37) for each block signal,
The modified differential signal subjected to the adaptive discrete cosine transform output from the output terminal (38) is transmitted to the still image transmission device on the partner side.

In the decoding system, the modified differential signal subjected to the adaptive discrete cosine transform (ADCT) from the input terminal (50) is subjected to inverse adaptive discrete cosine transform (IADCT) (51), and the modified differential signal (FIG. 9I) is obtained. This is the output terminal (4
9) and the digital luminance signal for one frame (composed of 768 × 480 pixel signals constituting a matrix of 8 bits each of 480 rows and 768 columns) (FIG. 9H) and added and filtered (52), and output Output to the terminal (53), enlarge the output and write it to the video memory (6) in FIG.
A corrected digital luminance signal (still image signal) (FIG. 9J) consisting of 768 × 480 pixel signals in 768 rows and 768 columns is obtained.

The digital luminance signal (still image signal) for one frame from the output terminal (53) is transmitted to the monitor receiver (8) shown in FIG.
, The corrected still image described above is further finely corrected in the still image transmission device of the other party. According to this correction, as described above, not only the roughness of the still image by interpolation but also the roughness by ADCT and IADCT can be corrected.

In the ADCTs (24), (30), and (37), the quantization matrix, DC Huffman code table, and AC Huffman code table stored in the external RAM (16) of the digital signal processing circuit (15) are as follows. Use the same thing.

[Problems to be solved by the invention]

In such a conventional still image transmission apparatus that performs hierarchical compression encoding and hierarchical expansion encoding, a digital still image signal for one frame is once subjected to low-pass filtering / decimation processing during hierarchical compression encoding. Divide for each block signal,
Each block signal is block-coded (first coding)
That is, while performing a discrete cosine transform (orthogonal transform),
Huffman coding (compression coding by variable length coding) is performed to obtain a compression-coded digital still image signal of a coarse still image, which is transmitted, and also subjected to low-pass filtering and decimation processing. The digital still image signal for one frame is interpolated, and the low-pass filtered / decimated digital still image signal is subtracted and filtered to obtain a modified difference signal, and the modified difference signal is block-coded. (Second block coding), that is, discrete cosine transform (orthogonal transform), and Huffman coding (compression coding by variable length coding) to obtain a compression-coded modified difference signal, This is transmitted.

Further, at the time of hierarchical expansion decoding, a digital still image signal of a coarse still image which has been compressed and encoded is expanded and decoded for each block, and interpolation processing is performed to obtain a digital still image signal of a coarse still image. After expanding and decoding the coded modified difference signal for each block signal, the digital still image signal of the coarse still image is modified without performing interpolation processing.

For this reason, the conventional still image transmission has a disadvantage that both the time required for the hierarchical compression encoding and the hierarchical expansion encoding and the time required for the transmission are both long.

Also, in the block coding used for the first and second coding, the same coding code table (block coding code table) has been conventionally used, so that it is the target of the first and second coding. The block coding for each signal could not be optimized. In particular, in the case of variable-length encoding in the first and second encoding, if the same variable-length encoding code table is used, the statistical properties of the signals to be subjected to the first and second variable-length encoding are changed. Because of the difference, the compression ratio is reduced.

In view of the above, the present invention provides an image transmission apparatus and an image transmission method in which an image signal is hierarchically encoded and transmitted, according to a difference between signals to be encoded in the first and second encodings. Further, the present invention intends to propose a device capable of performing optimal coding.

[Means for solving the problem]

According to a first aspect of the present invention, there is provided a first decimation unit that performs a first decimation process on an input image signal to generate a first decimation image signal, and a second decimation process on the first decimation image signal. A second thinning unit that performs processing to generate a second thinned image, and a first thinning unit that encodes the second thinned image signal using the first code table to generate first encoded data. Encoding means, decoding means for decoding the first encoded data by decoding processing corresponding to the encoding processing, and generating the decoded second thinned-out image signal;
Interpolation means for interpolating the thinned-out image signal to generate an interpolated image signal corresponding to the first thinned-out image signal, and calculating means for calculating the first thinned-out image signal and the interpolated image signal to generate a difference signal; A second encoding unit that encodes the differential signal using a second code table different from the first code table to generate second encoded data; and a first and a second encoded data. And a transmission unit for transmitting the image.

[Action]

According to the first aspect of the present invention, the first thinning unit performs the first thinning process on the input image signal to generate the first thinned image signal, and the second thinning unit generates the first thinned image signal. A second thinning image is generated by performing a second thinning process on the thinned image signal, and the second thinned image signal is encoded by the first encoding unit using the first code table. , Generating first encoded data, decoding the first encoded data by a decoding process corresponding to the encoding process by a decoding unit, generating a decoded second thinned image signal, and performing interpolation. Means for interpolating the decoded second thinned image signal to generate an interpolated image signal corresponding to the first thinned image signal,
The differential signal is generated by calculating the first thinned image signal and the interpolated image signal, and the differential signal is encoded by the second encoding means using a second code table different from the first code table. Then, the second encoded data is generated, and the transmission means transmits the first and second encoded data.

〔Example〕

Hereinafter, an embodiment of the still image transmitting apparatus according to the present invention will be described. Since the overall configuration is substantially the same as that of the still image transmitting apparatus shown in FIG. 6, the still image transmitting apparatus shown in FIG. An embodiment to which the hierarchical compression encoding and the hierarchical expansion decoding shown in the figure are applied will be described. In FIG. 1, parts corresponding to those in FIG. 8 are denoted by the same reference numerals and described.

Each memory in FIG. 1 is one of a plurality of areas of a buffer memory provided separately from the memories shown in FIG. 6, and the buffer memory is a bus (5) shown in FIG.
Shall be connected to However, the memory container of each memory is different from that of FIG.

Further, at the time of transmission and reception, the communication memory (14) of FIG. 6 temporarily stores a header signal for each frame of a digital luminance signal and two digital color difference signals, which will be described later, and a compression-encoded signal following the header signal. Used as memory for

First, compression coding by the hierarchical ADCT and decompression decoding by the hierarchical IADCT by the microcomputer (1) and the digital signal processing circuit (15) will be described mainly with reference to FIG.

The microcomputer (1) resets the digital signal processing circuit (15) at the beginning of each of the encoding process and the decoding process of the digital signal processing circuit (15), and resets the external memory of the digital signal processing circuit (15). In the encoding and decoding memory banks of (16), encoding and decoding programs, encoding quantization matrices, Huffman code tables (DC and AC Huffman code tables), zigzag scanning pointers are respectively stored in the encoding and decoding memory banks. And the like, as well as an inverse quantization matrix for decoding, an inverse Huffman code table (an inverse Huffman code table for DC and AC), and data for an inverse zigzag scanning pointer, respectively.

Also, under the control of the microcomputer (1), a digital luminance signal for one frame stored in the video memory (6) (768 × 480 pixel signals of 8 bits each forming a matrix of 480 rows and 768 columns). And one frame of digital red difference signals (each of which constitutes a matrix of 480 rows and 384 columns is composed of 384 × 480 pixel signals of 8 bits) and 1
A digital blue difference signal for a frame (each of which constitutes a matrix of 480 rows and 384 columns is composed of 384 × 480 pixel signals of 8 bits) is output for each digital luminance signal, digital red difference signal and digital blue difference signal. , Each of which is divided into block signals and output for each block signal.
When writing to M (15), during the exchange of intermediate data for the encoding and decoding quantization matrices, the Huffman code table, and the zigzag scanning pointer, and at the end of encoding and decoding for each block signal Also, the digital signal processing circuit (15) is reset.

Next, the hierarchical adaptive discrete cosine transform (hierarchical ADCT) and the hierarchical inverse adaptive discrete cosine transform (hierarchical IADCT) will be described with reference to FIG. 1. Here, the case of a still image signal as a luminance signal will be described. However, in the case of the red difference signal and the blue difference signal, the description thereof is omitted in principle.

In the encoding system (hierarchical compression encoding system) of FIG. 1, a digital luminance signal for one frame stored in a video memory (frame memory) (6) of FIG. 6 under the control of the microcomputer (1). (Still image signal) (This is a matrix of 480 rows and 768 columns, each of which is 8 bits 768 × 4
80 pixel signals) is divided into block signals composed of 32 × 32 adjacent pixel signals forming a matrix of 32 rows and 32 columns, which are read out for each block signal, and Supply to the input terminal (20) to perform low-pass filtering and 1/2 thinning processing (meaning two-dimensional low-pass filtering processing and the subsequent horizontal and vertical 1/2 thinning processing, and so on.) After (21), the data is reduced and written to the memory (22). Also, under the control of the microcomputer (1), a block signal composed of 16 × 16 adjacent pixel signals constituting a matrix of 16 rows and 16 columns written in the memory (22) is read out and subjected to low-pass filtering. A 1/2 thinning process (23) is performed to obtain a block signal composed of 8 × 8 adjacent pixel signals forming a matrix of 8 rows and 8 columns, and this is processed by an adaptive discrete cosine by a digital signal processing circuit (15). Convert (ADCT) (24).

This ADCT (24) is an 8 × 8 matrix forming an 8 × 8 matrix.
Each block signal composed of the pixel signals is written into an external RAM (16) of the digital signal processing circuit (15), and the digital signal processing circuit (15) performs two-dimensional discrete cosine transform (orthogonal transform). Thus, as shown in FIG. 7C, a DC coefficient signal (a signal of an average value of 8 × 8 coefficient signals) comes to the upper left corner, and from the low and high levels as the distance from the horizontal and vertical directions increases. The signal of the frequency coefficient will be distributed.

Then, each coefficient of the 8 × 8 coefficient signals is converted to an external RA.
Quantization is performed by dividing by 8 × 8 predetermined values corresponding to the divisor of the 8 × 8 quantization matrix stored in M (16), thereby obtaining 8 rows and 8 columns. 8 × 8 quantized coefficient signals forming an eight-column matrix are obtained. This 8x
Among the eight quantized coefficient signals, the quantized DC coefficient signal is located at the upper left corner, and many coefficient signals having a coefficient of 0 (for example, 2/3 of the whole) are distributed far from the upper left corner. ing.

In this manner, the DC coefficient signal at the upper left corner of the 8 × 8 coefficient signals forming the matrix of 8 rows and 8 columns that has been subjected to the two-dimensional discrete cosine transform and quantization may be subjected to Huffman coding as it is. Here, the difference between the coefficient of the previous block signal and the DC coefficient signal is taken, and the difference signal is stored in the external RAM (1
According to the DC Huffman code table written in 6),
Huffman coding (variable length coding) is performed.

Also, of the 8 × 8 coefficient signals, 8 excluding the DC coefficient signal is used.
× 8-1 coefficient signals are obtained by Huffman coding.
The zigzag scanning is performed by 8 × 8−1 zigzag scanning pointers forming an 8 × 8 matrix stored in the external RAM (16) so that the run length of 0 becomes longer and the compression ratio becomes higher. After that, 0 of the obtained 8 × 8-1 coefficient signals
The Huffman coding is performed based on the Huffman code table for AC stored in the external RAM (16) as a set of the run length and the non-zero value.

The Huffman coded coefficient signal for each block signal composed of 8 × 8 pixel signals obtained as described above is obtained by starting with the header signal created by the microcomputer (1), and then successively following it. The data is written into the communication memory (14) through the output terminal (31), is subjected to communication processing and modulation by the communication processing circuit / interface (12), and is transmitted through the transmission line (13) to the still image transmission device on the other side. You.

A Huffman coded coefficient signal corresponding to a block signal composed of 8 × 8 pixel signals subjected to adaptive discrete cosine transform (ADCT) from an input terminal (39) of a decoding system is converted into a digital signal processing circuit (15). By the inverse adaptive discrete cosine transform (IADCT) (40).

In the IADCT (40), a Huffman coding coefficient signal corresponding to a block signal composed of 8 × 8 pixel signals is converted into a DC and AC Huffman code table and an inverse zigzag stored in an external RAM (16). Each of the scanning pointers performs inverse Huffman coding to obtain a quantized coefficient signal forming a matrix of 8 rows and 8 columns, and a quantization matrix of 8 rows and 8 columns stored in the external RAM (16) is added thereto. Then, after multiplying by a corresponding predetermined multiplier and performing inverse quantization, two-dimensional inverse cosine transform is performed to obtain a block signal composed of original 8 × 8 pixel signals.

A block signal composed of 8 × 8 pixel signals obtained by the ADCT (24) is subjected to 2/1 interpolation processing (meaning 2/1 interpolation processing in each of horizontal and vertical directions) under the control of the microcomputer (1). (The same applies hereinafter.) (41), and then the data is enlarged and written into the memory (42). A block signal composed of 16 × 16 pixel signals constituting a matrix of 16 rows and 16 columns written in the memory (22) is read out from the block signal and subjected to 2/1 interpolation processing (43). Thus, a block signal composed of 32 × 32 pixel signals forming a matrix of 32 rows and 32 columns is obtained.

If a block signal composed of 32 × 32 pixel signals is sequentially written into the video memory (6) in FIG. 6, each of the 480 × 768 matrixes is composed of 8-bit 768 × 480 pixel signals. As a result, a digital luminance signal (still image signal) for one frame subjected to the interpolation processing is obtained.

If the interpolated digital luminance signal (still image signal) for one frame is reproduced by the monitor receiver (8) in FIG. 6, the still image can be seen promptly although it is a coarse still image. Can be.

Next, a first method for correcting the still image signal of the coarse still image is described.
The creation of the modified difference signal and its encoding and decoding will be described.

First, in the coding system, 8 × 8 block signals output to an output terminal (25) and subjected to adaptive discrete cosine transform (ADCT) (24) are inverse adaptive discrete cosine transformed by a digital signal processing circuit (15). Cosine transform (IA
DCT) (26) [same as the inverse adaptive discrete cosine transform (IADCT) (40) of the decoding system], thereby obtaining a block signal composed of the original 8 × 8 image signals, which is stored in the buffer memory. Written.

The block signal composed of 8 × 8 pixel signals stored in the buffer memory is subjected to 2/1 interpolation processing (27) [2/1 interpolation processing of decoding system (41) under the control of the microcomputer (1). )], A block signal composed of 16 × 16 pixel signals obtained by the enlargement processing is written to the memory (28).

FIG. 2A shows the contents stored in the memory (28), and a small rectangle indicates a block signal composed of 8 × 8 pixel signals. Then, this memory (28) has 2/1 interpolation processing (2
7) The block signal composed of 16 × 16 pixel signals constituting a matrix of 16 rows and 16 columns (this corresponds to four block signals composed of 8 × 8 pixel signals) is 15 Row 24
A block signal composed of 8 × 8 pixel signals is stored so as to form a matrix of rows, that is, a matrix of 30 rows and 48 columns.

FIG. 2B shows the stored contents of the memory corresponding to FIG. 2A in the case of the color difference signal.

The block signal composed of 16 × 16 pixel signals written in the memory (28) is stored in the memory (42) of the decoding system [the storage contents thereof are the same as those of the memory (28)]. This is subtracted and filtered (29) from the block signal composed of 16 × 16 pixel signals read from the memory (22), and is subjected to 2/1 interpolation processing (41) in the decoding system. ) To obtain a first modified difference signal (consisting of 8 × 8 difference pixel signals) for the block signal composed of 16 × 16 pixel signals.

The first modified differential signal is subjected to an adaptive discrete cosine transform (ADCT) (30) by a digital signal processing circuit (15), and the adaptive discrete cosine transformed modified differential signal output from an output terminal (31) is output. Is written in the communication memory (14) via the output terminal (31) sequentially from the header signal created by the microcomputer (1), and is sequentially processed by the communication processing circuit / interface (12). The signal is modulated and transmitted to the still image transmission device on the other side through the transmission line (13).

In the decoding system, the first modified difference signal subjected to the adaptive discrete cosine transform (ADCT) from the input terminal (45) by the digital signal processing circuit (15) is converted into the inverse adaptive discrete signal by the digital signal processing circuit (15). A cosine transform (IADCT) (46) is performed to obtain a first corrected difference signal, which is controlled by the microcomputer (1) from the 16 × 16 pixel signals read from the memory (42). After being filtered (47) with the block signal
It is subjected to 2/1 interpolation processing (48) and output to an output terminal (49). If the output is enlarged and written into the video memory (6) shown in FIG. 6, a matrix of 480 rows and 768 columns, each of which is composed of 8 bits of 768 × 480 pixel signals, corresponds to one corrected frame. Is obtained as a digital luminance signal (still image signal).

The DC and AC Huffman code tables and quantization matrices of the IADCT (46) stored in the external RAM (16) correspond to the DC and AC Huffman code tables and quantization matrices in the ADCT (30). And microcomputer (1)
Rewritten by

If the interpolated digital luminance signal (still image signal) for one frame is reproduced by the monitor receiver (8) in FIG. 6, the coarse still image described above will be corrected somewhat finely. This correction can correct not only the roughness of the still image by interpolation but also the roughness by ADCT and IADCT.

Next, the creation of a second modified difference signal for further modifying the still image signal of the still image that has been slightly finely modified, and the encoding and decoding thereof will be described.

First, in the encoding system, the adaptive modified cosine transform (ADCT) (30) first modified difference signal output to the output terminal (31) is subjected to inverse adaptive discrete cosine transform (IADCT) (32) [decoding Same as the inverse adaptive discrete cosine transform (IADCT) of the system (46), thereby returning the original first modified difference signal and the 16 × 16 signals read from the memory (28). And a block signal composed of the pixel signals of (1) and (3).

The result of the addition filtering (33) is subjected to a 2/1 interpolation process (3), and then enlarged and written to a memory (35).

When a third modified difference signal is also created, encoded, and decoded, the same processing as the creation, encoding, and decoding of the second modified difference signal is performed. Then, the still image signal obtained at the output terminal (53) may be corrected with the third corrected difference signal.

FIG. 3A shows the contents stored in the memory (35), and a small rectangle indicates a block signal composed of 8 × 8 pixel signals. Then, a block signal composed of 16 × 16 pixel signals constituting a matrix of 16 rows and 16 columns, which has been interpolated by 2/1 (34), is stored in the memory (35) (this is a block signal of 8 × 8 pixel signals). (Corresponding to four block signals) comprising a matrix of 30 columns and 48 rows, that is, a block signal composed of 8 × 8 pixel signals constitutes a matrix of 60 rows and 96 columns. Is stored.

This memory (35) needs the above-mentioned memory capacity when generating the third modified difference signal.
When the corrected difference signal is not created, a capacity capable of storing 192 block signals composed of 8 × 8 pixel signals is sufficient if only the 1/2 thinning process is considered. As described above, since 9 × 9 pixel signals are required for low-pass filtering of 8 × 8 pixel signals,
A capacity capable of storing 192 9 × 9 pixel signals is required. That is, a number of 0, 1,...
A block signal composed of × 8 pixel signals is stored in a memory (3
5), four block signals (for example, block signals of numbers 0, 1, 96, and 97) are used at a time. However, when ADCT (37) is performed, 8 × 8 pixel signals are used. , Respectively. Therefore, the numbers 0, 1,
This is because these block signals can be rewritten when the ADCT (37) of the block signals composed of 8 × 8 pixel signals of 2,..., 190, 191 is completed.

On the other hand, the contents stored in the memory (28)
Is used to generate the modified differential signal of the above, and as described above, a capacity for storing 48 × 30 block signals composed of 8 × 8 pixel signals is required.

FIG. 3B shows the contents stored in the memory corresponding to FIG. 3A in the case of the color difference signal.

The block signal composed of 16 × 16 pixel signals written in the memory (35) is read out and input to the input terminal (2
0), a block signal composed of 16 × 16 pixel signals obtained by subtraction filtering (36) from a block signal composed of 32 × 32 pixel signals and performing a 2/1 interpolation process (48) in a decoding system , A second corrected difference signal (consisting of 8 × 8 difference pixel signals) is obtained.

The second modified difference signal is subjected to adaptive discrete cosine transform (ADCT) (37) for each block signal by the digital signal processing circuit (15), and the output terminal (38)
The adaptive discrete cosine transformed first modified difference signal output from the communication memory (14) is output from the header signal created by the microcomputer (1) at the top, and then successively through the output terminal (38). The communication processing circuit / interface (12) performs communication processing and modulation, and transmits the communication processing circuit / interface (12) to the other party's still image transmission device via the transmission line (13).

In the decoding system, the adaptive modified cosine transform (ADCT) second modified difference signal from the input terminal (50) is
An inverse adaptive discrete cosine transform (IADCT) (51) is performed to obtain a second modified difference signal, which is output to an output terminal (4
9), ie, 16 from the video memory (6) in FIG.
Block signal consisting of × 16 pixel signals and addition filtering (4
7) is output to the output terminal (53), and the output is enlarged and written into the video memory (6) in FIG. 6 to obtain 768 × 480 pixel signals of 480 rows and 768 columns of 8 bits each. Thus, a digital luminance signal (still image signal) for one frame that has been subjected to the correction processing is obtained.

The digital luminance signal (still image signal) for one frame from the output terminal (53) is transmitted to the monitor receiver (8) shown in FIG.
, The corrected still image described above is further finely corrected in the still image transmission device of the other party. This correction can correct not only the roughness of the still image by interpolation but also the roughness by ADCT and IADCT.

Note that, after this, the third, fourth,... Corrected difference signals can be created to further finely correct the still image of the still image signal, but the second corrected difference signal is sufficient. Yes,
Even if it is increased, it is up to the third modified difference signal.

In addition, AD performed by the digital signal processing circuit (15)
The CT stores a quantization matrix, a DC Huffman code table, and an AC Huffman code table stored in its external RAM (16) in a video memory (6) for each of a luminance signal and a red and blue difference signal. The microcomputer (1) rewrites the still image digital video signal to an appropriate one according to the contents of the still image digital video signal for one frame.
Depending on whether the block signal composed of 8 × 8 pixel signals to be compression-coded by CT is a still image signal of a coarse still image or a modified difference signal, the degree of modification of the modified difference signal, That is, ADCT (24), (30), (37)
In this way, the microcomputer (1) rewrites an appropriate signal in consideration of the statistical properties of the signal to be encoded.

Note that the DC and AC Huffman code tables can be created in consideration of the statistical properties of the signals to be Huffman-coded, thereby increasing the compression ratio by Huffman coding. The same applies to general variable-length coding.

It is performed by the digital signal processing circuit (15).
IADCT (40), (26) and (46), (32) and (51),
A quantization matrix, a DC Huffman code table, and an AC Huffman code table stored in the external RAM (16) in (37), and the corresponding external RAM in ADCT (24), (30), and (37), respectively. It is rewritten according to the quantization matrix, DC Huffman code table, and AC Huffman code table stored in (16).

In the still image transmission apparatus of the above-described embodiment, for example, in the case of a digital luminance signal (still image signal), one frame of 768 × 480 pixel signals forming a matrix of 480 rows and 768 columns is stored in the frame memory. A digital luminance signal is stored and is divided into block signals (referred to as first block signals) forming a matrix of 32 rows and 32 columns.
As a result, this frame memory stores 24 × 15 first block signals constituting a matrix of 15 rows and 24 columns.

Then, for the 24 × 15 first block signals forming the matrix of 15 rows and 24 columns, from the left end of the first row to the right end, from the left end of the second row to the right end,... A low-pass filtering and 1/2 thinning-out process are sequentially performed from the left end to the right end of the 15th row, and a block signal (16th row of 16 × 16 pixel signals forming a matrix of 16 rows and 16 columns) (second Block signals), and each of them was sequentially written into a memory (referred to as a first memory) so as to form a matrix of 15 rows and 24 columns, and read from the first memory. Each time the first block signal is stored in the first memory, it is subjected to low-pass filtering and 1/2 thinning-out processing to form 8 × 8 pieces of matrix each having 8 rows and 8 columns. A block signal (third block signal) composed of pixel signals is obtained and written into a memory (referred to as a second memory). I have to.

By the way, low-pass filtering for each pixel signal constituting a block signal is performed by dividing a certain pixel signal into a certain pixel signal and eight surrounding pixel signals, that is, a weighted average of the levels of a total of nine pixel signals. Means that the pixel signal is replaced with a pixel signal having a level of Therefore, in order to perform low-pass filtering on the second block signal, 17 × 17 pixel signals forming a matrix with 17 rows and 17 columns are required.

Therefore, in order to perform a low-pass filtering process on the second block signal at the upper left corner of the first memory to obtain the third block signal, the upper and left sides of the second block signal and the third block signal are obtained. 16 + 16 + 1 = 33 pixel signals are virtually provided outside the corner, and the level of the virtual pixel signal is the middle level between the minimum and maximum levels of the pixel signal, or 16 + 16−1 = 31 inside the pixel signal. What is necessary is just to employ | adopt this pixel signal.

In order to perform low-pass filtering on the second block signal on the right of the second block signal at the upper left corner of the first memory, the upper side and the corner of the second block signal are required. A virtual 17 pixel signal is provided on the outside of the unit, and the level of the virtual pixel signal is the middle level between the minimum and maximum levels of the pixel signal, or the lower and upper left corners thereof.
The levels of 16 + 1 = 17 pixel signals may be employed, and the eight pixel signals in one column on the right side of the second block signal at the upper left corner of the first memory may be used. Further, for each of the second block signals in the first row following this, low-pass filtering may be performed in the same manner.

In order to perform low-pass filtering on a second block signal immediately below the second block signal at the upper left corner of the first memory, the left block of the second block signal and the corner thereof are required. Outside of the pixel signal, 17 virtual pixel signals are provided, and the level of the virtual pixel signal is set to the middle level between the minimum and maximum levels of the pixel signal, or 16 levels at the right and upper left corners thereof.
+ 1 = 17 pixel signal levels and the first
In this case, eight pixel signals in one row below the second block signal at the upper left corner of the memory may be used. In addition, the first
For each second block signal in the column, low-pass filtering may be performed in a similar manner.

Also, for 24 × 15 second block signals constituting a matrix of 15 rows and 24 columns stored in the first memory, from the left end of the first row to the right end, from the left end of the second row to the right end, ...
・ Low-pass filtering in order from left to right on line 15
A 1/2 thinning-out process is performed to obtain a third block signal composed of 8 × 8 pixel signals forming a matrix of 8 rows and 8 columns, and each of them is sequentially stored in the second memory in 15 rows and 24 columns. When performing low-pass filtering on each second block signal except for the second block signal in the first row on the upper side and the first column on the left side when writing to form the matrix of As shown in FIG. 4, 16 × 16 pixel signals constituting the matrix of 16 rows and 16 columns are added to the second block signals of the upper, left and upper left corners adjacent to the second block signal. , 16 + 16 + 1 =
A low-pass filtering process is performed on 17 × 17 pixel signals forming a matrix of 17 rows and 17 columns obtained by adding 33 pixel signals.

In addition, as shown in FIG.
The second block signal composed of × 16 pixel signals has 32 rows.
First consisting of 32 × 32 pixel signals forming a matrix of 32 columns
32 + 32 + 1 on the left, upper and upper left corners of the block signal
= 65 pixel signals added, obtained by referring to 33 × 33 pixel signals forming a matrix of 33 rows and 33 columns, 16 rows
Second consisting of 16 × 16 pixel signals forming a 16-column matrix
16 + 16 + 1 on the left, upper and upper left corners of the block signal
= 17 pixel signals forming a matrix of 17 rows and 17 columns to which 33 pixel signals are added are the left side of the first block signal,
It can be seen that the result is obtained by referring to 35 × 35 pixel signals forming a matrix of 35 rows and 35 columns in which 32 + 32 + 4 = 68 pixel signals are added to the upper and upper left corners.

Next, referring to FIG. 5, low-pass filtering / 1/2 decimation processing and 2/1 interpolation processing in the encoding system and decoding system of FIG. 1 and low-pass filtering / 1/2 processing in FIG. A supplementary explanation will be given on the thinning process. In FIG. 5, for simplicity,
The first block signal is converted to an 8 × 8 matrix forming an 8 × 8 matrix.
It is composed of eight pixel signals, and the second block signal is 4
The third block signal is constituted by 2 × 2 pixel signals constituting a matrix of 2 rows and 2 columns, and the third block signal is constituted by 4 × 4 pixel signals constituting a matrix of 4 rows and 4 columns.

In FIG. 5A, 8 + 8 +
9 × 9 pixel signals forming a matrix of 9 rows and 9 columns, with 1 = 17 pixel signals added (small rectangles indicated by broken lines)
On the other hand, low-pass filtering and 1/2 decimation processing are performed, and a second block signal (4 × 4 pixel signals forming a matrix of 4 rows and 4 columns) (a small rectangle indicated by a solid line, The inside is filled with dots).

In FIG. 5B, 5 × 5 pixel signals constituting a matrix of 5 rows and 5 columns (4 × 4 + 1 = 13) are added to the second block signal obtained in FIG. The small rectangle shown in FIG. 2 is filled with dots, and low-pass filtering and 1/2 thinning processing are performed to obtain a third block signal (2 × 2 matrix forming a 2 × 2 matrix). (Two pixel signals) (small rectangles shown by solid lines, the inside of which is filled with oblique lines). Then, the third block signal in the lower right part of the boundary line a is subjected to the ADCT processing.

In FIG. 5C, a 3 × 2 pixel signal (solid line) forming a matrix of 2 rows and 2 columns constituting a third block signal obtained by IADCT processing is obtained by converting the third block signal subjected to ADCT processing. And six pixel signals of a second block signal above and on the left side of the third block signal (a small rectangle indicated by a broken line, and (The interior is filled with dots and dashed slashes) in the horizontal and vertical directions,
Interpolating the pixel signal of the level of the average value of the level of the pair of pixel signals (a small rectangle indicated by a solid line, the inside of which is filled with dots) between the pair of pixel signals, (A small rectangle shown by a solid line, the inside of which is filled with oblique lines and dots). Then, the second block signal at the lower right portion of the boundary line b is interpolated by 2/1 in FIG. 5D.

In FIG. 5D, 4 × 4 pixel signals forming a matrix of 4 rows and 4 columns (a small rectangle indicated by a solid line and a dot inside the matrix) constituting the second block signal obtained in FIG. And the second block signal on the left side of the second block signal.
Using the four pixel signals of the block signal (a small rectangle shown by a broken line, and the inside is filled with dots), the pair of pixel signals is disposed between the pair of pixel signals in the horizontal and vertical directions. A third block signal (a small rectangle indicated by a solid line) is interpolated by interpolating a pixel signal at the level of the average value of the pixel signals (a small rectangle indicated by a solid line and the inside is filled with dots). , Dot-filled and unfilled).

〔The invention's effect〕

According to the first aspect of the present invention, a first decimation unit that performs a first decimation process on an input image signal to generate a first decimation image signal, and a second decimation unit that generates a first decimation image signal
A second thinning-out process is performed to generate a second thinned-out image.
Thinning means, first coding means for coding a second thinned image signal using a first code table to generate first coded data, and coding the first coded data Decoding means for decoding by a decoding process corresponding to the process to generate the decoded second thinned-out image signal; and interpolating the decoded second thinned-out image signal to correspond to the first thinned-out image signal. Interpolating means for generating an interpolated image signal;
Calculating means for calculating a thinned image signal and an interpolated image signal to generate a difference signal; and coding the difference signal by using a second code table different from the first code table to perform second coding. Second encoding means for generating data; first and second encoding means;
And a transmission means for transmitting the coded data of (a), (b), and (c), a code table suitable for each of the image signals (the original image signal or the difference signal) can be used. A transmission device can be obtained.

According to a second aspect of the present invention, in the image transmission apparatus according to the first aspect of the present invention, the first encoding means generates cosine data by performing cosine transform on the second thinned image signal for each block.
The coefficient data is quantized to generate quantized data, and the quantized data is variable-length coded using a first code table to generate first coded data.
The second encoding means generates cosine data by cosine transforming the difference signal for each block, quantizes the coefficient data to generate quantized data, and converts the quantized data into a first code table. Since it is configured to generate second encoded data by performing variable length encoding using a different second code table, an image signal (original image signal or difference signal) is generated.
Therefore, it is possible to obtain an image transmission device that can perform optimal variable-length coding.

According to the third aspect of the present invention, the thinning-out processing means for performing the thinning-out processing on the input image signal to generate the thinned-out image signal, and encoding the thinned-out image signal by using the first code table, thereby obtaining the first A first encoding unit that generates encoded data, a decoding unit that decodes the first encoded data by a decoding process corresponding to the encoding process, and generates a decoded decimated image signal; Interpolating means for interpolating the thinned-out image signal to generate an interpolated image signal corresponding to the input image signal; calculating an input image signal and an interpolated image signal of the interpolating means to generate a difference signal; Is encoded using a second code table different from the
And a transmission means for transmitting the first and second encoded data, so that a code table suitable for an image signal (an original image signal or a difference signal) is provided. Can be used, whereby an image transmission device that can perform optimal encoding can be obtained.

According to a fourth aspect of the present invention, in the image transmission apparatus according to the third aspect of the present invention, the first encoding means generates cosine data by cosine transforming the thinned-out image signal for each block, and quantizes the coefficient data. To generate quantized data, and the quantized data is subjected to variable-length encoding using the first code table,
The second encoding unit is configured to generate first encoded data, and the second encoding unit generates cosine data by cosine transforming the difference signal for each block, and quantizes the coefficient data to generate quantized data. Then, the quantized data is variable-length coded using a second code table different from the first code table to generate second coded data. It is possible to use a code table suitable for each of the image signal and the difference signal), thereby obtaining an image transmission device capable of performing optimal variable-length coding.

According to the fifth aspect of the present invention, the step of performing the first decimation process on the input image signal to generate the first decimation image signal, and the second decimation process on the first decimation image signal Generating a second decimated image by applying
Encoding a second thinned image signal using a first code table to generate first encoded data;
Decoding the encoded data by a decoding process corresponding to the encoding process to generate a decoded second thinned-out image signal, and interpolating the decoded second thinned-out image signal to generate a first thinned-out image signal. Generating an interpolated image signal corresponding to the decimated image signal; calculating the first decimated image signal and the interpolated image signal to generate a difference signal; 2 and a step of transmitting the first and second encoded data by encoding using the code table No. 2 and transmitting the first and second encoded data. ), You can use the appropriate code table for each,
As a result, it is possible to obtain an image transmission method capable of performing optimal encoding.

According to a sixth aspect of the present invention, in the image transmission method according to the fifth aspect of the invention, in the step of generating the first encoded data, the second thinned image signal is subjected to cosine transform for each block to generate coefficient data. Then, the coefficient data is quantized to generate quantized data, and the quantized data is subjected to variable-length encoding using a first code table to generate first encoded data, and to perform second encoding. In the step of generating data,
The difference signal is cosine-transformed for each block to generate coefficient data, the coefficient data is quantized to generate quantized data, and the quantized data is converted to a second code table different from the first code table. Variable-length coding to generate the second coded data, it is possible to use a code table suitable for each of the image signals (original image signal or difference signal). An image transmission method capable of performing long coding can be obtained.

According to the seventh aspect of the present invention, a step of performing a decimation process on an input image signal to generate a decimation image signal, and encoding the decimation image signal using the first code table to generate a first code Generating decoded data, decoding the first encoded data by a decoding process corresponding to the encoding process, generating a decoded thinned image signal, and interpolating the decoded thinned image signal. Generating an interpolated image signal corresponding to the input video signal by
A difference signal is generated by calculating the input image signal and the interpolated image signal of the interpolation means, and the difference signal is encoded using a second code table different from the first code table to form a second code. Generating the encoded data and transmitting the first and second encoded data, it is possible to use a code table suitable for the image signal (original image signal or difference signal), respectively. Accordingly, it is possible to obtain an image transmission method capable of performing optimal encoding.

According to an eighth aspect of the present invention, in the image transmission method according to the seventh aspect of the present invention, in the step of generating the first coded data, the thinned image signal is subjected to cosine transform for each block to generate coefficient data. Quantize coefficient data to generate quantized data, perform variable length coding on the quantized data using a first code table, generate first coded data, and generate second coded data And cosine transforming the difference signal for each block to generate coefficient data, quantize the coefficient data to generate quantized data, and convert the quantized data to a second code table different from the first code table.
Since the second encoded data is generated by performing variable-length encoding using the code table of (1), a code table suitable for each of the image signals (the original image signal or the difference signal) can be used. Accordingly, it is possible to obtain an image transmission method capable of performing optimal variable-length coding.

According to the ninth aspect of the present invention, in an image decoding device that generates an image signal by decoding an encoded signal including the received first and second encoded data, the first and second encoded data are Receiving means for receiving, first decoding means for decoding the encoded data by using the first code table to generate a thinned image signal, and decoding the second encoded data to a first code table different from the first code table. A second decoding means for decoding using the code table of No. 2 to generate a differential signal, and a first decoding means for interpolating the thinned-out image signal.
And an adder for adding the first image signal and the difference signal to generate a second image signal, so that a code table suitable for each encoded signal is used. Accordingly, it is possible to obtain an image decoding device capable of performing optimal decoding.

According to a tenth aspect of the present invention, in the image decoding apparatus according to the ninth aspect of the present invention, the first decoding means performs variable length decoding of the first encoded data using the first code table to obtain quantized data. Is generated, and the quantized data is inversely quantized to generate coefficient data, and the coefficient data is subjected to inverse cosine transform to generate a thinned-out image signal. Variable-length decoding the quantized data using a second code table different from the first code table to generate quantized data;
Dequantize the quantized data to generate coefficient data,
Since the differential data is generated by performing inverse cosine transform on the coefficient data, it is possible to use a code table suitable for each of the coded signals, thereby performing optimal variable-length decoding. An image decoding device capable of performing the above can be obtained.

According to an eleventh aspect of the present invention, in the image receiving method for generating an image signal by decoding an encoded signal including the received first and second encoded data, the first and second encoded data are Receiving, decoding the encoded data using the first code table to generate a thinned image signal, and converting the second encoded data to a second code table different from the first code table. Generating a differential signal by decoding using the same, interpolating the thinned image signal to generate a first image signal, adding the first image signal and the differential signal to generate a second image signal. Generating a code table suitable for each coded signal, thereby obtaining an image receiving method capable of performing optimal decoding.

According to the twelfth aspect of the present invention, in the image receiving method according to the eleventh aspect of the present invention, in the step of generating a thinned image signal, the first encoded data is subjected to variable length decoding using the first code table to perform quantum decoding. Generating quantized data, inversely quantizing the quantized data to generate coefficient data, inverse cosine transforming the coefficient data to generate a thinned image signal, and generating a difference signal. The data is variable-length decoded using a second code table different from the first code table to generate quantized data, the quantized data is inversely quantized to generate coefficient data, and the coefficient data is inversely decoded. Since a differential signal is generated by performing cosine transform, it is possible to use a code table suitable for each of the encoded signals, thereby obtaining an image receiving method capable of performing optimal variable-length decoding. In Wear.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of hierarchical compression encoding and hierarchical decompression decoding in an embodiment of the present invention, FIG. 2 is an explanatory diagram of encoding in the embodiment, FIG. 3 is an explanatory diagram of decoding in the embodiment, FIG. FIG. 4 is an explanatory diagram of low-pass filtering / decimating processing of the embodiment, FIG. 5 is an explanatory diagram of low-pass filtering / decimating processing and interpolation of the embodiment, and FIG. 6 is a block diagram showing a conventional still image transmission apparatus. FIG. 7 is an explanatory diagram of conventional encoding, FIG. 8 is a block diagram showing conventional hierarchical coding and hierarchical decoding, and FIG. 9 is a conventional hierarchical coding and hierarchical decoding. FIG. (1) a microcomputer, (6) a video memory (frame memory), (12) a communication processing circuit / interface, (13) a transmission line, (14) a communication memory,
(16) is its external RAM, (17) is its internal RAM, (21),
(23) is low-pass filtering, 1/2 decimation, (27), (3
4), (41), (43), and (48) are 2/1 interpolation processing, (2
9), (36) are subtraction filtering, (33), (47), (52)
Is an addition filtering process.

Continuation of the front page (56) References JP-A-63-306789 (JP, A) JP-A-63-73786 (JP, A) JP-A-2-16887 (JP, A) IEICE Technical Report, IE 88-80 (1988) p. 63-70 IEEE TRANSACTIONS ON COMMUNICATIONS Vol. COM-31 [4] (1983. 4) p. 532-540 (58) Fields surveyed (Int.Cl. ⁶ , DB name) H04N 7/24-7/68 H04N 1/41-1/419

Claims (12)

(57) [Claims] A first thinning means for performing a first thinning process on an input image signal to generate a first thinned image signal; and a second thinning process for the first thinned image signal. And a second thinning-out means for generating a second thinned-out image by applying a first code table to the second thinned-out image signal to generate first coded data. Encoding means; decoding means for decoding the first encoded data by a decoding process corresponding to the encoding process to generate the decoded second thinned-out image signal; Interpolating means for interpolating the thinned-out image signal to generate an interpolated image signal corresponding to the first thinned-out image signal, and calculating the first thinned-out image signal and the interpolated image signal to generate a difference signal Calculating means; A second encoding unit that encodes using a second code table different from the code table to generate second encoded data; and a transmission unit that transmits the first and second encoded data. An image transmission device comprising: 2. The image transmission apparatus according to claim 1, wherein said first encoding means generates cosine data by performing cosine transform on said second thinned-out image signal for each block. The second code is configured to generate quantized data by quantization, and to perform variable length coding on the quantized data using the first code table to generate the first coded data. A cosine transform unit for generating the coefficient data by cosine transforming the difference signal for each block, quantizing the coefficient data to generate quantized data, and converting the quantized data to a first code table different from the first code table. 2. An image transmission apparatus characterized in that it is configured to perform variable length encoding using the code table of No. 2 and generate the second encoded data. 3. A thinning processing means for performing a thinning process on an input image signal to generate a thinned image signal, and encoding the thinned image signal using a first code table to form first encoded data. A decoding unit that decodes the first encoded data by a decoding process corresponding to the encoding process, and generates a decoded thinned-out image signal. Interpolating means for interpolating the thinned-out image signal to generate an interpolated image signal corresponding to the input image signal; calculating the input image signal and the interpolated image signal of the interpolating means to generate a difference signal; Is encoded using a second code table different from the first code table,
An image transmission device comprising: a second encoding unit that generates second encoded data; and a transmission unit that transmits the first and second encoded data. 4. The image transmission apparatus according to claim 3, wherein said first encoding means generates cosine data by cosine transforming said thinned-out image signal for each block, and quantizes said coefficient data. It is configured to generate quantized data, perform variable length coding on the quantized data using the first code table, and generate the first coded data, and the second coding means Generating cosine data by cosine transforming the difference signal for each block, quantizing the coefficient data to generate quantized data, and converting the quantized data to the second code table different from the first code table. An image transmission apparatus configured to perform variable-length encoding using a code table to generate the second encoded data. 5. A step of performing a first decimation process on an input image signal to generate a first decimation image signal; and performing a second decimation process on the first decimation image signal. Generating a second decimated image; encoding the second decimated image signal using a first code table to generate first encoded data; Decoding by a decoding process corresponding to the encoding process to generate the decoded second decimated image signal; and interpolating the decoded second decimated image signal, the first decimated image Generating an interpolated image signal corresponding to the signal, generating the differential signal by calculating the first thinned image signal and the interpolated image signal, and converting the differential signal into the first code table. Different second And encoded using code tables, and generating a second coded data, image transmission method characterized by comprising the step of transmitting the first and second encoded data. 6. The image transmission method according to claim 5, wherein, in the step of generating the first encoded data, the second thinned image signal is subjected to cosine transform for each block to generate coefficient data. The coefficient data is quantized to generate quantized data, and the quantized data is subjected to variable-length coding using the first code table to generate the first coded data. In the step of generating quantized data, the difference signal is subjected to cosine transform for each block to generate coefficient data, the coefficient data is quantized to generate quantized data, and the quantized data is converted to the first code table. An image transmission method, wherein variable-length encoding is performed using a second code table different from the above, to generate the second encoded data. 7. A step of performing a decimation process on an input image signal to generate a decimation image signal, encoding the decimation image signal using a first code table, and generating first encoded data. Generating, decoding the first encoded data by a decoding process corresponding to the encoding process, and generating a decoded decimated image signal; and interpolating the decoded decimated image signal. Generating an interpolated image signal corresponding to the input image signal, and generating a differential signal by calculating the input image signal and the interpolated image signal of the interpolating means. Is encoded using a second code table different from
An image transmission method, comprising: generating second encoded data; and transmitting the first and second encoded data. 8. The image transmission method according to claim 7, wherein in the step of generating the first encoded data, the thinned image signal is subjected to cosine transform for each block to generate coefficient data. Is quantized to generate quantized data, and the quantized data is subjected to variable-length coding using the first code table to generate the first coded data. The second coded data is In the generating step, the difference signal is cosine-transformed for each block to generate coefficient data, the coefficient data is quantized to generate quantized data, and the quantized data is different from the first code table. The variable length coding is performed using the second code table,
An image transmission method characterized in that encoded data is generated. 9. An image decoding apparatus for decoding an encoded signal comprising received first and second encoded data to generate an image signal, comprising: receiving the first and second encoded data; Means, first decoding means for decoding the encoded data using a first code table to generate a thinned image signal, and decoding the second encoded data different from the first code table. A second decoding unit that decodes using the second code table to generate a difference signal; an interpolation unit that generates a first image signal by interpolating the thinned image signal; An image decoding apparatus comprising: an adding unit configured to add the difference signal to generate a second image signal. 10. The image decoding apparatus according to claim 9, wherein said first decoding means performs variable length decoding of said first encoded data by using said first code table to convert quantized data. And generating the coefficient data by inversely quantizing the quantized data, and generating the thinned image signal by inverse cosine transform of the coefficient data. Variable-length decoding of the encoded data using the second code table different from the first code table to generate quantized data, inverse quantization of the quantized data to generate coefficient data, An image decoding apparatus configured to generate a difference signal by performing inverse cosine transform of coefficient data. 11. An image receiving method for generating an image signal by decoding an encoded signal composed of received first and second encoded data, wherein the first and second encoded data are received. Decoding the coded data using a first code table to generate a thinned image signal; and converting the second coded data into a second code table different from the first code table. Generating a differential signal by decoding using the same; generating the first image signal by interpolating the thinned image signal; adding a second image by adding the first image signal and the differential signal; Generating an image signal. 12. The image receiving method according to claim 11, wherein the step of generating the thinned-out image signal includes the step of:
Is subjected to variable-length decoding using the first code table to generate quantized data, inversely quantized the quantized data to generate coefficient data, and inverse cosine-transformed the coefficient data. In the step of generating a thinned-out image signal and generating the difference signal, the second encoded data is subjected to variable-length decoding using the second code table different from the first code table to obtain quantized data. Produces
An image receiving method, wherein the quantized data is inversely quantized to generate coefficient data, and the coefficient data is inversely cosine transformed to generate a difference signal.