JP3016790B2 - Image encoding device and decoding device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an encoding device suitable for encoding an image signal such as a color moving image including a still image, and an encoding device using the encoding device. The present invention relates to a decoding device that converts the encoded code into an image signal.

[Prior Art] Conventionally, as this kind of technique, there is one disclosed in the following document.

"" Kosugi, Matsuda, Sakai: A Study of Filters in Motion Compensated Cosine Transcoding Loop "IEICE Technical Report IE86-100" Figures 2 (a) and 2 (b) 1 is a block diagram illustrating a configuration of an image encoding device and a decoding device disclosed in the above document.

 The encoding device will be described with reference to FIG.

The original image is converted into an electric signal representing the light and shade distribution by scanning the imaging tube of the TV camera 201, and then the A / D converter (A
/ D) 202 sampled into pixels and converted to digital image signals. Then, this digital image signal is
Further, color separation is performed by a color separation circuit (Y / C) 203 to obtain a luminance signal and a color difference signal.

The luminance signal / color signal is a motion vector detector (MV)
The value of m is compared with the contents of the frame memory 205 by
A motion vector is calculated for each size processing block (for example, 16 × 16) of × m pixels, and motion compensation is performed using this vector to obtain a predicted value.

Next, the prediction value motion-compensated with the motion vector previously obtained by the adder 206 is subtracted, and an inter-frame difference residual signal is obtained. At this time, the calculated motion vector is encoded and transmitted to the transmission path 209.

On the other hand, the inter-frame difference residual signal is orthogonally transformed by a discrete cosine exchanger (DCT) 207 in units of n × n processing blocks (for example, 8 × 8), and is decorrelated. At this time, the transform coefficients in the orthogonal transform are quantized by a quantizer (Q) 208, converted into a quantization level number, and transmitted to the transmission path 209.

The output of the quantizer 208 is also connected to an inverse quantizer (Q ^-1 ) 210, and the quantization level number is converted into an inverse quantum value by the inverse quantizer 210, and further, an inverse discrete cosine The signal is restored by the converter (DCT ^-1 ) 211 into an inter-frame difference residual signal.

The predicted value subtracted by the adder 206 is added to the restored inter-frame difference residual signal by the adder 212, and the result is input to the frame memory 205 as a local decoder reproduced signal.

Next, the decoding device will be described with reference to FIG.

First, the quantization level number and the motion vector of the residual signal are received from the transmission path 213.

The quantization level number is inversely quantized by an inverse quantizer 214, and the inversely quantized value is converted by an inverse discrete cosine transformer 215 to restore an inter-frame difference residual signal.

Next, a predictor predicted by the motion vector received from the transmission path 213 is added to the restored inter-frame difference residual signal by an adder 216, and the local decoder on the encoder side is added. It is converted into a reproduction signal exactly the same as the reproduction signal.

This reproduced signal is transmitted to the frame memory 217 and the color synthesis circuit 2
Entered in 18. The color synthesizing circuit (Y / C) 218 synthesizes a luminance signal and a color difference signal into a digital image signal. The synthesized digital image signal is converted into an analog image signal by the D / A converter 219 and reproduced by the image monitor 220.

[Problems to be Solved by the Invention] However, the above-described conventional method has the following problems.

Since quantization and inverse quantization are performed in units of blocks, if the characteristics of the quantizer are different in each block, the difference in the quantization error appears as block distortion of the image.

The high frequency component of the orthogonal transform coefficient is lost by quantization,
The outline of the image may be blurred.

The present invention has been made under the above-mentioned background,
It is an object of the present invention to provide an image encoding device and a decoding device that do not cause block distortion or blur in an outline.

[Means for Solving the Problems] The present invention has solved the above-described problems by adopting the following configurations.

(1) Image region dividing means for dividing an image according to the characteristics of each pixel of a digital image signal, and coding means for coding the outer periphery of each image region divided by the image region dividing means. An image encoding apparatus for converting original image information into a digital image signal and encoding the digital image signal, wherein the image area dividing means comprises: a separating means for dividing the digital image signal into a luminance signal and a color difference signal; and a luminance signal and a color difference signal. And a codebook having a plurality of representative vectors that are regarded as components of a vector, and comparing a vector composed of the divided luminance signal and color difference signal with a plurality of representative vectors of the codebook,
Storage means for storing the number of the closest representative vector, wherein the image is divided into regions by classifying and representing each pixel according to the stored number of the representative vector. .

(2) In the configuration 1, the encoding unit has a function of erasing a pixel in the area where the encoding is completed, and a function of erasing a pixel even if the area has a cavity due to the erasure of the pixel. A configuration having a function of encoding only the outer periphery.

(3) In the configuration 1, the encoding unit may track an outer periphery of an image area having the same representative vector number, set a tracking posture (direction) at the time of the tracking to one state, and change the tracking posture. And a function of a virtual tracking moving object (robot) for encoding by expressing the state transition by a code as a state transition.

(4) In the configuration 3, the encoding unit detects the end of encoding of each of the image areas by monitoring a tracking start point, a current position being tracked, and a code generation state of the tracking moving object. A configuration characterized by having functions.

(5) decoding means for inputting the encoded code and decoding the outer periphery of each image area divided by the image encoding device, and a digitized image signal before being divided from the decoded image information 2. An image decoding apparatus for inputting and decoding a code encoded by the image encoding apparatus according to claim 1, further comprising: an image reproducing unit configured to reproduce the image data. A storage unit for storing a vector number, a codebook for outputting a vector having a luminance signal and a color difference signal as constituent elements corresponding to the representative vector numbers stored in the storage unit, and outputting as a vector from the codebook And a color synthesizing means for performing color synthesis based on the obtained luminance signal and color difference signal.

(6) In the configuration 5, even when the decoded area has a cavity, the decoding means has a function of decoding only the outer periphery of the area. A configuration in which a function of reproducing a pixel having the same property as the outer periphery of the completed area is provided.

(7) In the configuration 5, the decoding means tracks the outer periphery of the image area having the same representative vector number, sets the tracking attitude (direction) at the time of the tracking to one state, and changes the tracking attitude. And a function of a virtual tracking moving object (robot) for decoding by expressing the state transition by a code as a state transition.

(8) In the configuration 7, the decoding unit monitors a tracking start point of the tracking moving object, a current position being tracked, and a code generation state to detect decoding end of each image area. A configuration characterized by having functions.

[Operation] According to the above configuration 1, the image area dividing unit performs the area division of the image by expressing the property of each pixel by a vector and classifying each pixel according to the value of the vector. Since the function is provided, the area can be divided relatively easily.

According to the configuration 2, the encoding unit performs a function of erasing a pixel in the area where the encoding is completed, and a function of erasing the pixel even if the area has a cavity due to the erasure of the pixel. Since only the function of encoding the inner circumference is provided, efficient encoding can be performed without erroneously encoding the inner circumference.

According to the configuration 3, the encoding unit tracks the outer periphery of the image area under a certain rule, sets a tracking posture (direction) at the time of the tracking to one state, and determines a change in the tracking posture. Since a virtual tracking moving object (robot) function is provided for encoding by expressing the state transition as a code as a state transition, efficient encoding using an extremely simple codeword is possible.

According to the fourth aspect, the encoding means monitors the tracking start point, the current position being tracked, and the code generation status of the tracking moving object, and detects the end of encoding of each of the image areas. Since the function is provided, it is possible to detect the end of encoding without requiring any special means for detecting the end of encoding.

According to Configurations 5 to 8, it is possible to input and decode a code coded by the coding apparatus according to Configuration 1, and reproduce the original image.

Embodiment FIG. 1 is a block diagram showing a configuration of an image encoding device and a decoding device according to an embodiment of the present invention, and FIG. 1 (a) shows an image encoding device. FIG. 1 (b) shows the decoding device. One embodiment described in detail below shows an example in which the present invention is applied to a color moving picture coding apparatus and a decoding apparatus.

(Encoding Device) First, an image encoding device will be described based on FIG.

In the figure, reference numeral 101 denotes a TV camera (TV), and reference numeral 102 denotes A /
D converter, code 103 is a color separation circuit (Y / C), code 104 is a vector quantizer (VQ), code 105 is an image memory (FM), code 106 is a boundary tracking coding circuit (C), code 107 Is a transmission path.

The image signal input from the TV camera 101 is output to the A / D converter 10
It is sampled by 2 and becomes a digital image signal.

The digital image signal is separated by the color separation circuit 103 into a luminance signal Y and two color difference signals C ₁ and C ₂ .

Next, these three types of signals are input to the vector quantizer 104.

The vector quantizer 104 regards each pixel value of the signals Y, C ₁ and C ₂ as an element of the three-dimensional vector X, and performs vector quantization according to the following procedure based on a codebook prepared in advance. .

A luminance signal Y of a certain pixel I (x, y) and color difference signals C ₁ and C ₂
From these, a three-dimensional vector X = (Y, C ₁ , C ₂ ) is formed.

The vector X and the representative vector V ₁ to
Calculate the distance of V _N by brute force.

If the closest representative vector is _VK , the pixel I (x,
y) is given a vector number k.

 Is performed for all pixels.

The vector numbers thus obtained are input to the image memory 105 and stored.

The vector numbers stored in the image memory 105 are read out and encoded by the boundary tracking encoding circuit 106 and sent to the transmission path 107.

Hereinafter, the contents of the image memory 105 and the operation of the boundary tracking encoding circuit 106 will be described in detail.

FIG. 3 is an explanatory diagram of the storage state of the vector numbers stored in the image memory. As shown in FIG. 3, the vector numbers are distributed so as to form an area for each number, reflecting the properties of the input screen (brightness, color, etc. of the subject). The boundary tracking coding circuit 106 performs coding by tracking and following the outer periphery of this area.

The encoding process by the boundary tracking encoding circuit 106 is roughly performed in the following four processes.

1) Detection of the m-th area A _Km of the vector number k.

2) Tracking coding of the outer periphery of the area A _Km .

3) Elimination of area A _Km .

4) Return to 1) with m = m + 1. If there is no area with the vector number k, k = k + 1, m = 1, and a code indicating that encoding of all the areas with the vector number k has been completed is sent, and the procedure returns to 1).

FIG. 4A is a block diagram showing an internal configuration of the boundary tracking encoding circuit 106. In the figure, reference numeral 401 denotes an area detector (RD), and reference numeral 402 denotes a tracking simulator (Si).
m), reference numeral 403 is a counter. Hereinafter, with reference to FIG. 4 (a), among the above-described processes 1) to 4),
The processes 1) to 3) will be described in detail.

Process 1) The area detection circuit 401 scans the contents of the image memory 105 and searches for an area A _km having a vector number k. If you discover
A start signal is sent to the tracking simulator 402, and the process proceeds to the process 2).

2) Process The tracking simulator 402 tracks the outer periphery of the area A _km on the image memory 105 by performing a tracking simulation by a virtual tracking robot as shown in FIG.

This virtual tracking robot tracks a vector number based on the following rules.

* The right hand of the robot must be on the area A _km with the vector number k.

* The left hand of the robot must be outside the area A _km .

Doing simulated tracking robot by these two rules,該追traces robot keeps one of the tracking position of the four tracking position as shown in the state transition diagram of FIG. 6 (state S ₀ to S ₃₎ While tracking the vector numbers one after another. The code “0” representing the tracking posture at this time
~ "2" is output. Code “0” to “2” representing this tracking posture
Is obtained by encoding the area A _km .

Hereinafter, the operation of the tracking simulator 402 will be described in more detail.

First, when receiving the start signal from the area detector 401, then the encoded addresses of the pixels area detector 401 has been scanned as the start address A _S region encoded and transmitted to a transmission line 107.

As an initial state S _0, put a hand tracking robot to the start address A _S.

Next, the tracking robot includes the following (a) to (c)
Is performed, and the place where the right hand is newly placed is set as a new address _Aj .

(A) When the pixel p ₁ = k, the left hand is rotated left 90 ° about the axis to output a code “1”, and the state transits to S _{mod (n + 3, 4)} . here,
“Mod (n + 3,4)” represents the remainder of (n + 3) divided by 4.

(B) when the pixel p ₁ ≠ k and pixel p ₂ = k, out the code "0" to move forward, to continue the S _n.

(C) When the pixel p ₁ ≠ k and the pixel p ₂ ≠ k, 90
゜ Rotate to output code “2” and transit to S _{mod (n + 1,4)} .

FIGS. 7A to 7C are operation explanatory diagrams corresponding to the above operations (a) to (c), respectively.

Codes “0” to “2” output in the above operation are transmission lines.
It is sent to 107.

At this time, when the code "2" is output, the counter 403 is incremented by one.

Further, when a code "0" or "1" is output, the counter 403 is reset to zero.

Next, the pixel value 0 is written to the old address A _j-1 of the image memory 105 because it has been encoded.

 The above is repeated.

As a result of the above operation, when A _j = A _s (when the region has made one round) or when the counter 403 has reached 4 (when the region having one pixel has made one round), the region A _{Assuming that} the _km encoding has been completed, the process proceeds to the next process 3).

Process 3) In process 3), as shown in FIG. 8, when one region A _km includes regions of several different vector numbers, the inner periphery of A _km is erroneously track-encoded. In order not to erase A _km . Hereinafter, the procedure will be described in detail.

From the upper left corner of the actual image data to the lower right,
The pixel value of the vector number k that is in contact with the pixel having the pixel value of 0, which is a mark that has been coded on the image memory 105, is newly set to 0.

 The same operation is performed from the upper right corner to the lower left.

 The same operation is performed from the lower right corner to the upper left.

 The same operation is performed from the lower left corner to the upper right.

By performing the above operations (1) to (4), the area A _km whose outer periphery is coded in the process 2) is stored in the image memory 1.
Deleted from 05. Thereby, only the outer periphery is encoded one after another without erroneously tracking and encoding the inner periphery, so that waste can be reduced.

(Decoding Device) Next, an image decoding device will be described based on FIG. 1 (b).

In the figure, reference numeral 108 denotes a transmission path, reference numeral 109 denotes a boundary tracking decoding circuit (D), reference numeral 110 denotes an image memory (FM), and reference numeral 111
Denotes a vector inverse quantizer (VQ ^-1 ), reference numeral 112 denotes a color synthesis circuit (Y / C), reference numeral 113 denotes a D / A converter (D / A), and reference numeral 114 denotes an image monitor.

The code from the transmission line 108 is input to a boundary tracking decoding circuit 109.
And converted into a vector number and stored in the image memory 110. The stored vector number is read out by the dequantizer 111 and dequantized into a (Y, C ₁ , C ₂ ) vector according to the codebook. Next, these Y, C ₁ , C ₂ are
Are converted into analog image signals by the D / A converter 113 and reproduced by the image monitor 114.

Hereinafter, the boundary tracking decoding circuit 109 will be described in detail.

In short, the boundary tracking decoding circuit 109 reproduces the area of the vector number on the image memory 110 by following the reverse process of the above-described boundary tracking encoding circuit 106. I do.

1) vector number k, receives a start address A _s of the m-th region A _miles.

2) Tracking decoding of the outer periphery of the area A _km is performed.

3) Return to the above 1) with m = m + 1. If a code indicating that the encoding of all the regions of the vector number k has been completed is received, k = k + 1 and m = m + 1 and the process returns to the above 1).

4) Fill the vector number k inside the area A _km .

The above processes 1) to 4) are repeated to restore the vector numbers one after another.

FIG. 4 (b) is a block diagram showing an internal configuration of the boundary tracking decoding circuit 109. In the figure, reference numeral 404 denotes a tracking simulator (Sim), and reference numeral 405 denotes a controller (Co).
nt) and reference numeral 406 are counters (Count). Hereinafter, the fourth
Each of the above processes 1) to 4) with reference to FIG.
Will be described in detail.

Process 1) The controller 405 receives the start address information and sends a start signal to the tracking simulator 404 to start tracking decoding.

Process 2) The tracking simulator 404 performs the following operation to reproduce the outer periphery of the area of the vector number k.

First, receives a start signal from the controller 405, the start address A _s a region encoding the transmission line
Receive from 108.

The initial state as S _0, put a hand tracking robot to the start address A _s.

Next, the tracking robot includes the following (a) to (c)
Is performed, and the place where the right hand is newly placed is set as a new address _Aj . At this time, when the code "2" appears, the counter 406 is incremented by one. Further, when the code "0" or "1" is output, the counter 406 is reset to zero.

Upon receipt of the (a) reference numeral "1", considered pixel p ₁ = k, by rotating left by 90 ° to the axis of the left hand, a transition to S _{mod (n + 3,4).}

(B) When the code “0” is received, the pixel p ₁ ≠ k and the pixel p ₂ =
thought k, to continue the S _n moving forward.

(C) When the code “2” is received, the pixel p ₁ {k and the pixel p ₂ }
Think k, rotate your right hand 90 ° about its axis, and S _{mod (n + 1,4)}
Transitions to.

FIGS. 7A to 7C are operation explanatory diagrams corresponding to the above operations (a) to (c), respectively.

Next, the pixel value k is written to the old address A _j-1 of the image memory 105 because it has been decoded.

 The above is repeated.

3) Process In process 3), A _j = A _s (when the area has made one round) or the counter 406 becomes 4 and the area A
_The controller 405 detects that the encoding of _km has been completed, and resets the tracking simulator 404 and the counter 406. Thereafter, returning to the process 1), the tracking decoding of the new area A _{km + 1} is performed.

When a code indicating that the encoding of the area of the vector number k has been completed is received, the tracking simulator 404,
The counter 406 is reset, and thereafter, the process returns to the process 1) to perform tracking decoding of the new area A _{k + 11} .

When the end of decoding for all vector numbers and all areas is detected, the process proceeds to the next process 4).

4) Process In process 4), the area A surrounded by the vector number k
_A vector number k is filled in _{km by} the following procedure.

From the upper left corner of the actual image data to the lower right,
A pixel having a pixel value of 0 which is in contact with a pixel having a pixel value k which is a decoded mark on the image memory 110 is newly set as a pixel value k.

 The same operation is performed from the upper right corner to the lower left.

 The same operation is performed from the lower right corner to the upper left.

 The same operation is performed from the lower left corner to the upper right.

By performing the above operations (1) to (4), the area A _km whose outer periphery has been decoded in the process 2) is stored in the image memory 1.
10 is filled with a vector number k.

In this way, the image transmitted from the transmission line 108 is reproduced on the image memory 110 with the vector number.

Thereafter, the vector dequantizer 111 converts the signal into a luminance signal Y and color difference signals C ₁ and C _2.
12. A reproduced image is obtained through the D / A converter 113 and the image monitor 114.

(Advantages of One Embodiment) According to the above-described embodiment, the following advantages can be obtained.

Think of an image as a collection of areas, similar colors,
Since the outer periphery of the region having brightness is encoded and transmitted, block-like distortion does not occur.

Considering a virtual tracking robot that tracks the outer periphery of the area to encode the area, and changes the tracking posture as a state transition and encodes it, so "0" ~
Tracking data can be composed of three codewords “2”. That is, it can be constituted by a short code of 1 to 2 bits.

Considering a virtual tracking robot that tracks the outer periphery of the area to encode the area, and changes the tracking posture as a state transition and encodes it, the end of encoding of one area is the start address. The code "2" can be automatically detected when it appears four times in succession, and no special code indicating the end of encoding is required.

After the area is coded, the area is erased, so that the inner circumference of the area is not erroneously coded and coding can be performed efficiently.

It should be noted that the present invention is not limited to the above-described embodiment, and it is needless to say that the present invention can be applied to an encoding device and a decoding device for monochrome moving images or color or monochrome still images.

[Effects of the Invention] As described above in detail, the present invention relates to an image encoding apparatus for converting original image information into a digital image signal and encoding the digital image signal. A configuration including image area dividing means for performing division, and encoding means for encoding the outer periphery of each image area divided by the image area dividing means; and inputting a code encoded by the encoding apparatus. An image decoding apparatus for decoding the image area; decoding means for decoding the outer periphery of each image area divided by the encoding apparatus; and a digital image signal before being divided from the decoded image information. And an image encoding apparatus that does not cause block distortion or blur in contours. And a decoding device.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention, FIG. 2 is a block diagram showing the configuration of a conventional image encoding device and a conventional decoding device, and FIG. 3 is a vector number stored in an image memory. FIG. 4 is a block diagram showing an internal configuration of a boundary tracking encoding circuit and a boundary tracking decoding circuit, FIG. 5 is an operation explanatory diagram of a virtual tracking robot, and FIG. 6 is a state transition diagram. , FIG. 7 is an explanatory diagram of a tracking operation by a virtual tracking robot, and FIG. 8 is an explanatory diagram of region coding. 101 TV camera (TV), 102 A / D converter, 103 Color separation circuit (Y / C), 104 Vector quantizer (VQ), 10
5 ... Image memory (FM), 106 ... Boundary tracking coding circuit (C), 107 ... Transmission path, 108 ... Transmission path, 109 ... Boundary tracking decoding circuit (D), 110 ... Image memory (FM), 111
... Vector inverse quantizer (VQ ^-1 ), 112... Color synthesis circuit (Y / C), 113... D / A converter (D / A), 114.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N 7/133 Z

Claims (8)

(57) [Claims] An image area dividing means for dividing an image according to the characteristics of each pixel of a digital image signal, and an encoding means for encoding an outer periphery of each image area divided by said image area dividing means. An image coding apparatus for converting original image information into a digital image signal and encoding the digital image signal, wherein the image area dividing unit includes: a separating unit that divides the digital image signal into a luminance signal and a color difference signal; A codebook having a plurality of representative vectors that regard the color difference signal as a component of the vector, and a vector composed of the divided luminance signal and color difference signal is compared with the plurality of representative vectors of the codebook. Storage means for storing numbers, and by classifying and representing each pixel according to the stored number of the representative vector, Image encoding device and performs image segmentation. 2. The coding means has a function of erasing a pixel in an area where the coding has been completed, and a function of erasing a pixel even if the area has a cavity due to the erasure of the pixel. 2. The image encoding apparatus according to claim 1, further comprising a function of encoding. 3. The encoding means tracks the outer periphery of an image area having the same representative vector number, sets a tracking posture (direction) at the time of the tracking to one state, and changes the tracking posture in a state. 2. The image encoding apparatus according to claim 1, further comprising a function of a virtual tracking moving body (robot) for encoding by expressing the state transition as a transition by a code. 4. An encoding end detecting function for detecting the end of encoding of each image area by monitoring a tracking start point, a current position being traced, and a code generation state of the tracking moving object. The image encoding device according to claim 3, further comprising: 5. A decoding means for receiving an encoded code and decoding the outer periphery of each image area divided by an image encoding device, and a digital signal before being divided from the decoded image information. An image decoding device for inputting and decoding a code encoded by the image encoding device according to claim 1, further comprising: an image reproducing unit that reproduces an image signal, wherein the image reproducing unit includes: A storage unit for storing the number of the representative vector, a codebook for outputting a vector having a luminance signal and a color difference signal as constituent elements corresponding to the number of the representative vector stored in the storage unit, a vector from the codebook An image decoding apparatus comprising: a color synthesizing unit that performs color synthesis based on a luminance signal and a color difference signal output as a color signal. 6. The decoding means has a function of decoding only the outer periphery of the decoded area even if the decoded area has a cavity, and the decoding is completed. 6. The image decoding apparatus according to claim 5, wherein the image decoding apparatus has a function of reproducing a pixel having the same property as an outer periphery of the area. 7. The decoding means tracks the outer periphery of an image area having the same representative vector number, sets a tracking posture (direction) at the time of the tracking to one state, and changes the tracking posture in a state. 6. The image decoding apparatus according to claim 5, further comprising a function of a virtual tracking moving object (robot) for decoding by representing the state transition by a code as a transition. 8. A decoding end detecting function for detecting the end of decoding of each image area by monitoring the tracking start point of the tracking moving object, the current position being tracked, and the code generation status. The image encoding device according to claim 7, further comprising: