JPH0697462B2 - Image rotation method - Google Patents

Description

The present invention relates to a method of rotating an image having a word structure, and is capable of processing one memory access unit in units of words. The present invention relates to an image rotation method capable of performing high-speed rotation processing on an image at any angle without causing distortion.

[Conventional technology]

As a conventional image rotation device, for example, Japanese Patent Laid-Open No. 55-95145
"Shape rotation method", JP-A-57-89171 "Image processing device", Information Processing Society of Japan 25th National Convention (Late 1982) Takeda et al. "Image tilt correction method by two-dimensional block transfer",
The Information Processing Society of Japan 26th National Convention (first half of 1983), Takeda et al. "High-speed image rotation method by two-dimensional block transfer", etc.

Conventionally, a method of performing an affine transformation is generally used for rotating a two-dimensional pattern. The method using this affine transformation is to perform coordinate transformation by the following equation.

However, x and y indicate the old coordinate system, X and Y indicate the new coordinate system, and θ indicates the rotation angle.

Each pixel position coordinate (x,
y) is substituted into the equation (1) to obtain the converted new coordinate (X, Y), and the pixel value at the old coordinate (x, y) is converted to the new coordinate (X, Y).
Rotation conversion is performed by substituting Y). If the transformed new coordinates (X, Y) are not both integers, that is, if they are not on the grid point, the lattice point (X ₀ , Y ₀ ) closest to (X, Y) is newly converted to the old coordinate (x, Y). A rotation transformation is performed by replacing the pixel value in y) with the new coordinate (X ₀ , Y ₀ ). Details of the equation (1) are described, for example, in pages 137 to 138 of Algebra and Geometry (authors Shuzo Wada and Fumiyuki Terada) published by Hirokawa Shoten.

As a specific example, there is a known rotary coordinate generation circuit as one of the graphic processing functions provided in the graphic display. It is composed of a sine cosine calculation circuit, a matrix register group, and a matrix multiplication circuit. Based on sin θ, cos θ obtained from the sine cosine calculation circuit by inputting the rotation angle θ, the matrix in equation (1) Are registered in the matrix register group. Next, the old coordinates (x, y) of the two-dimensional pattern and the contents of the matrix register can be input to the matrix multiplication circuit, and the new coordinates (X, Y) can be obtained as an output.

However, in this method, the calculation of the formula (1) has to be performed for each pixel of the two-dimensional pattern using the multiplier, the calculation time becomes long, and the rotating two-dimensional pattern cannot be obtained at high speed. . In addition, there is a problem that the circuit becomes complicated.

In order to solve this, in JP-A-55-95145, in a method of obtaining a rotated two-dimensional pattern by rotating a stored two-dimensional pattern by an arbitrary angle, the stored two-dimensional pattern is stored in the horizontal direction. Scanning at a predetermined angle, sequentially moving in the vertical direction to perform the first reading, and scanning in the horizontal direction, moving sequentially in the vertical direction to store and reconstruct the first reading signal, Scanning is performed at an angle perpendicular to the angled scanning, sequentially moving in the horizontal direction to perform the second reading, and scanning in the vertical direction, sequentially moving in the horizontal direction to obtain the second reading signal. A method of obtaining a rotating two-dimensional pattern by performing memory reconstruction has been proposed.

Further, in Japanese Patent Application Laid-Open No. 57-89171, there is provided an image processing device for obtaining image data of an image obtained by rotating a certain image, the image data inputting means for inputting the image data of the image, the image Image data encoding means for encoding and compressing image data input by the data input means, rotation data input means for inputting rotation center data and rotation angle data for the rotation, and
It is configured to include coded data processing means for processing the coded image data based on the input rotation data and outputting the image data of the rotated image, and performs diagonal scanning to perform multiplication. An apparatus has been proposed which obtains a rotation pattern without performing.

[Problems to be solved by the invention]

However, in the case of the conventional image rotation device, in the case of the reference 1, if the inclination is large, the image distortion occurs when the rotation angle becomes large because it extends over a plurality of words. Therefore, the rotation angle is in a minute range. Since the address is limited and it is necessary to control the address in bit units, there is a problem that the range of use is narrow and the configuration is complicated. Further, in the case of the reference 2 (Japanese Patent Laid-Open No. 57-89171), it is necessary to encode and decode the image in the rotation processing, and all dots must be processed even in the word structure. The configuration becomes complicated and the processing time becomes long. In addition, depending on the image content, the amount of data may increase when the image is encoded, and there is a disadvantage that the processing time increases. Also, references 3 and 4
Corresponds to the improved version of Reference 1, but since the processing is controlled in 1-bit units, there is a problem that the configuration of the address control unit becomes complicated.

[Means and Actions for Solving Problems]

The present invention has been made in view of the above, and can perform processing in word units without requiring multiplication processing and bit processing,
In addition, in order to rotate the image at an arbitrary angle at high speed without causing distortion, the image data is shifted by the angle θ by giving the rotation angle θ, and the shifted image data is determined by the rotation angle θ. After enlarging in the sub-scanning direction at the enlargement ratio, enlarge the image data to 90
And rotate the main scanning direction and sub-scanning direction,
An image rotation method is provided in which image data rotated by 90 degrees is reduced at a reduction rate determined by a rotation angle θ, and then the reduced image data is shifted by an angle θ to perform oblique axis exchange.

〔Example〕

 Hereinafter, the image rotation device according to the present invention will be described in detail.

Before describing the embodiments of the present invention, the principle of the present invention will be described.

As described above, the coordinate conversion of rotation can be expressed as in equation (1). Equation (1) shows the case where the figure is fixed and the coordinate axes are converted, but in order to make it easier to understand in the following description, the coordinate axes are fixed as in the original Cartesian coordinate system and the figure is converted and described. To do. First, set the point (X ₀ , Y ₀ ) to θ
The rotated point (X ₁ , X ₁ ) is expressed as follows.

The range of the angle θ is −45 ° θ45 °.

In particular, when | θ |> 45 °, it can be realized by combining the 90 ° rotation circuit and the processing method of the present invention. That is, the total rotation angle is θ _T, and the rotation angle according to the present invention is θ (−45 ° θ
45 °), θ _T = 1/90 + θ (where l = 0, ± 1, ...) Can be realized. Therefore, in the following explanation, -45 ° θ
Handle θ in the range of 45 °.

Next, as shown in FIG. 3, tan (-.theta.) Conversion of the oblique axis is performed line by line (that is, tan .theta. Is added line by line depending on the position in the sub-scanning direction), and only the result is shifted. (Here, the counterclockwise θ is positive).

Next, 1 / cos θ is sequentially added to the figure shifted as shown in FIG. 4 according to the equation (5), and when the value A carries, one line is inserted and the y-axis (sub scanning) Expand in the direction.

In order to shift the enlarged figure vertically,
Rotate 90 ° as shown in Fig. 5.

Here, as shown in FIG. 6, the figure rotated 90 ° is multiplied by cos θ in the y-axis (sub-scanning) direction, and the original figure is reduced in the x-direction by thinning the lines. Reduction ratios are sequentially added, and thinning is performed when no carry occurs.

Next, the reduced figure is subjected to oblique axis conversion by tan (-.theta.) As shown in FIG. 7 (processing for returning the conversion in FIG. 3 to the original).

The image is rotated by (90 ° -θ) by the ectopic processing.

Further, the oblique-axis-transformed figure is rotated by -90 degrees (or +270 degrees) as shown in FIG. 8 (processing for returning 90 degrees rotation in FIG. 5).

Here, (4), (5), (6), (7), (8),
From equation (9), That is, Then, the oblique axis conversion and the enlargement / reduction 90 ° rotation are performed to rotate the arbitrary angle.

The 90 ° rotation and -90 ° rotation may be interchanged in order.

At this time, the matrix operation is as follows.

Then, the oblique axis conversion, enlargement, -90 ° rotation, reduction, oblique axis conversion, 90 ° rotation can be performed at any angle.

The description has been made for the case of θ <0, but the same applies to the case of θ> 0. Further, in the above description, the respective processes are all performed separately, but in reality, it is possible to perform the oblique axis conversion and enlargement, the reduction, and the oblique axis conversion at the same time. That is,
Equations (4) and (5) are Equations (7) and (8) are Can be replaced and processed.

FIG. 2 shows an example of an editing apparatus to which the present invention is applied. An image input apparatus 1 which reads an original such as a document using a CCD or the like and outputs the image signal as a binary digital signal, The CPU 2 which is the core of the apparatus and executes the editing processing according to the set program, the image apparatus 3 which performs the rotation processing on the image data stored in the image memory by the processing method according to the present invention, and the processing apparatus during and after the processing. CRT4 for displaying images and keyboard 5 for inputting various commands
Between the image memories 6a and 6b for storing the image data input from the image input device 1 and the image data after the rotation process, the image output device 7 for hard-copying the edited image subjected to the rotation process, and the respective devices. It is composed of a data bus 8 which is connected to perform data transfer.

FIG. 1 shows an embodiment of the present invention and corresponds to the image rotation device 3 of FIG. The image rotation device 3 according to the present invention includes image memories 6a and 6b and an image shift unit 3 described later.
b and an image shift unit 3b for shifting the image in the line (main scanning) direction in the range of 0 to (W-1) bits, and a control unit 3a for controlling the address and read / write of the image 90 ° rotation unit 3c. , An image 90 ° rotating unit 3c for rotating the image by 90 °, and executes an oblique axis conversion process, an enlarging / reducing process and a 90 ° rotating process to rotate the image by an arbitrary angle.

FIG. 9 shows the details of the image shift unit 3b. The register 10 for setting the shift number, the data register 11 for setting the data to be shifted, the cyclic shift circuit 12 for circulating the data by the set number of bits, and the shift A data latch 13 that latches the overflow bit to the next data until it is combined with the next data, a data selector 14 that selects the previously overflowed data and the data shifted without overflowing this time into one word, and the data selector 14 And a bus driver 16 for outputting the shifted data. The registers 10 and 11 are configured by using a latch circuit or the like, and latch data by a write signal of the image processor or the like.

FIG. 10 shows the configuration of the cyclic shift circuit 12, which is composed of four sets of select circuits 12a, 12b, 12c, 12d. In this configuration, the shift is made to the right. The signals S _{0 to} S ₃ are the outputs of the shift number setting register 10, S ₀ is the LSB, and S ₃ is the MSB. In other words, the value of S _{0 to} S ₃ indicates a binary shift number. If S ₀ = L, the 1-bit shift unit does not shift, and if S ₀ = H, 1-bit shift is performed. That is, Si
If (i = 0 to 3) is L, the A input of the selector is output to Y, and if it is H, the B input is output. In the case of 6-bit shift, S ₀ = L, S ₁ = H, S ₂ = H, S ₃ = L, and the data propagates through the cyclic shift circuit 12, as shown in FIG.
A data latch circuit of the data latch circuit 13 is 16 bits (or a fact 15 bits. May in this case L ₁₅ ~L _1.), At the time of initial can clear latch data.
The latch timing is when the image processor finishes reading data from the image shift circuit.

Next, the data selector 14 will be described. Fig. 12 shows the configuration. 16 2-input selectors (15 can be used) are arranged in parallel. The input R is the output of the cyclic shift circuit 12, the input L is the output of the latch circuit, and the outputs M _{15 to} M ₀ are connected to the output bus driver. C _{15 to} C ₀ input to the SEL input are supplied from the selector control circuit 15. When the SEL input = L, the A input (R) is selected, and when the SEL input = H, the B input (L) is selected.

Next, the selector control circuit 15 will be described. This circuit can be realized by a gate, a ROM, or the like. Fig. 13 shows the case of implementation in ROM. The contents of the ROM are shown in the table below. The input signals S _{3 to} S ₀ are supplied from the shift number setting circuit. That is, the number of selecting latch data increases according to the number of shifts. Also, as you can see from the ROM output, since C ₀ is always 0, one selector in FIG. 12 is unnecessary. That is, R ₀ = M ₀ . Therefore, the LSB of the latch circuit and the selector control circuit 15 is not necessary, and each bit can be configured with 15 bits.

The operation in FIG. 9 will be described with reference to FIG. 14th
In the figure, a case is shown in which one word is shifted by 6 bits by 3 bits to the right. Original 1 line data 1
When a word is input, the data that is shifted by the set number of bits is output. The shifted and overflowed data is latched. When the next one word is input, this is also shifted by the set bit. Here, the previously latched data and the data that did not overflow due to the shift are selected and output as one word. The overflow is latched for the next word. Thereafter, such an operation is repeated, and the input data is sequentially shifted and output by the set number of bits. As a result, one line is shifted by the specified number of bits. Address control is performed when there is a shift of one word or more or another line.

Next, the image 90 ° rotating portion 3c will be described in detail.

Figures 15 and 16 (a) to (e) are image 90 ° rotating parts.
3c is shown in detail, FIG. 15 shows the structure of the circuit body, and parts 16 (a) to (e) show the contents stored in the image memories 6a and 6b.

First, the general operation of the image 90 ° rotating unit 3c will be described. W words are continuously shifted into the shift register (there are W shift registers) corresponding to each bit of the word. The shift-in direction is input from the LSB side to the MSB side of the shift register according to + 90 ° rotation or −90 ° rotation. At the place where the W word is shifted in, parallel reading is repeated W times from W shift registers in order according to the rotation direction. In order to rotate the entire image by 90 °, the resulting rotated image of W bits × W words may be transferred to the address where the block should be located after rotation. This may be done for the entire image.

Next, the circuit will be described in detail with reference to FIGS. 15 and 16 (a) to (e). As is clear from FIG. 16, the image memories 6a, 6
The image stored in b is composed of 8 × 8 pixels, and FIG. 16 (a) shows this as a video image, and the letter R drawn in black on a white background can be read. FIG. 9B is image data displayed as a binary code group by replacing the white part with 0 and the black part with 1 in FIG. 9B, which is the storage content itself of the image memory 3, and It can be regarded as a matrix of 8 rows and 8 columns. The first of this matrix
The second, third, ..., 8B rows are numbered 1B,
If they are 2B, 3B, ..., 8B, each of them constitutes one byte of data, and the content thereof is specified by the data D7, D6, D5 ,. If you display this in the form of a row matrix, Can be written. As will be described later, the image memories 6a and 6b
The reading of the data from or the writing of the data to the image memory 6 is performed with the data of each row (in this case, 1-byte data) as one unit. It can be said that each of the numerical sequences obtained by vertically reading the data in FIG. 16B from the left column constitutes a column matrix.

20a to 20h in FIG. 15 are 8-bit 8-state 3-state shift registers, and the data shift direction can be switched to the left shift or the right shift according to the designation from the CPU 2 as described later. The 8 × 8 image register 20 is arranged side by side, and the image data (binary code group) stored therein also forms a matrix of 8 rows and 8 columns. 21 is a decoder for selecting any one of the shift registers 20a to 20h based on the memory address signal sent from the CPU 2 via the address bus 100, and 110 is also issued from the CPU 2 to the image register 20. Address signal that transmits write signal WR, read signal RD, and chip selection signal CS issued to the CS terminal on the input side of decoder 21, 22 is only when chip selection signal CS is input to CS terminal of decoder 21 , A gate circuit that plays a role of transmitting the write signal WR toward the shift registers 20a to 20h.

The address bus 100 is the line 100a, 100b, 100c and 100d.
4a, of which 100a, 100b, 100c transmit memory address signals composed of 3-bit parallel binary code and are connected to the A, B and C terminals on the input side of the decoder 21 respectively. . Also, 100d is for the shift registers 20a-20h.
Connected in parallel to all L / R terminals, each shift register 2
0a to 20h function as a shift register for left shift when the signal Aφ transmitted thereby is 1 and as a shift register for right shift when they are 0. Reference numeral 8 denotes a data bus for transmitting image data exchanged between the image memory 6 and the image register 20, which is composed of 8 lines 9a to 9h as shown in FIG. 17 and has an input terminal of each shift register 20a to 20h ( DL and DR terminals) are connected to the lines 8a to 8h, and the output terminals Oφ to O7 are connected to the lines 8a to 8h, respectively.

Next, the function of each chip 21 and 20a to 20h will be described. First, the shift register 20x shown in FIG. 17 (x is a ...
Image data is input as parallel binary code from the image memory 6 through the line 8x in the data bus 8 to the input terminals DL and DR of any one of h) when the shift register 20x is left-shifted. Are taken into the shift register 20x from the DL terminal and from the DR terminal in the case of right shift. Further, D'φ to D'7 are sist registers 20
The data stored in x are output as parallel binary codes to the lines 8a to 8h in the data bus 8 through the output terminals O'φ to O'7, respectively. CK is an input terminal for a clock signal, and is an input terminal for a chip selection signal CS (a signal that selects this line) and a write signal WR. The CE terminal is called a chip enable terminal, and the shift register 20x responds to the read signal RD only when the signal 1 is input from the output terminal Ox of the decoder 21 to this terminal.
It outputs to the lines 8a to 8f in the data bus 8 through O'φ to O'7 of the shift register 20x. So in this case CE
It is safe to assume that the terminal functions similarly to the chip select terminal. The function of the L / R terminal has already been described.
As shown in FIG. 17, when the data shift register direction of the shift register 20x is left shift, the data is shifted to the left side (arrow direction in the same figure) as viewed from the shift register 20x itself, and the data is shifted to the left side in FIG. In the mounting state as shown in Fig. 17, the reader shifts downward.

Next, the A, B, C terminals on the input side of the decoder 21 are connected to the address bus 1
It is an input terminal for a memory address signal sent as a 3-bit parallel binary code through three lines 100a, 100b, and 100c in 00, and its output terminal O depends on the content of the input signal.
The signal 1 is output only from any one terminal of φ to O7, and the output from the other output terminals is 0. G ₁ and G ₂ are input terminals for the read signal RD and the chip selection signal CS (signal for selecting the chip 21), respectively.

In the above configuration, first the line 1 in the address bus 100
The signal Aφ = 1 is input through 00d and is transmitted to the L / R terminals of the shift registers 20a to 20h, so that all the shift registers function as the left shift register.

Next, as shown in FIG. 16B, the data 1B [0111100] of the first row of the image data stored in the image memory 6 is stored.
0] is read and input terminals (DL and DR terminals) of the shift registers 20a to 20h through the lines 8a to 8h in the data bus 8.
Input in parallel. Next, the write signal WR synchronized with the clock signal is transmitted through the control bus 11 and input to the CK terminals of the shift registers 20a to 20h through the gate circuit 22. Therefore, the signal (0 in this case) transmitted through the line 8a is taken into the inside through the input terminal DL of the shift register 20h and written as data D'φ. This state is shown in FIG. 18 (a). Next, by the same operation, the data in the image memory 6 (data 2B [01000100] in the second row in FIG. 16 (b) is read out, and the shift register
Data D'φ is written in parallel to 20a to 20h.
Since the data 1B already written shifts to the left (in the direction of the arrow in FIGS. 15 and 17), the state of the data array in the image register 20 is as shown in FIG. 18 (b). Below, the data in the image memory 6 (part 16 (b))
3rd, 4th, ..., 8th row are read out in the order (3B, 4B, ..., 8B order) and written into the shift registers 20a to 20h. The status of the data array is as shown in Figure 16 (c).

Here, the data arranged in the image register 20 (FIG. 16 (c)) is regarded again as a matrix of 8 rows and 8 columns.
The data of the 2nd, ..., 7th and 8th columns are respectively 8B ', 7B',
.., 2B ', 1B', each of them constitutes 1-byte data, the contents of which are the data D'shown in FIG. 16 (C).
.phi., D'1, ..., D'6, D'7 are specified and stored in the shift registers 20a to 20h, respectively. If you display this in the form of a column matrix, Next, the memory address signal 000 is input to the A, B, C terminals on the input side of the decoder 21 and the read signal RD is input to the G ₁ terminal on the input side of the decoder 21, and the signal 1 is output from the Oφ terminal on the output side. It will be output and will function as the CE pin of the shift register 20h. At this time, the other shift registers 20a to 20g do not function. The data in the shift register 20h (that is, the 16th
The data 1B ′ “00000000” in the eighth column of FIG. 3C is read out as parallel binary code from the output terminals O′φ, O′1, O′2, ... It is transmitted to the image memory 6 through each of the lines 8a to 8h of No. 8 and written as the data 1B of the first row here (the first row in FIG. 16 (d)). After that, the data of the 7th, 6th, 5th, ...
2B ', 3B', 4B ', ..., 8B' are successively shifted to the shift register 20.
The data 2B, 3B, 4B, of the second, third, fourth, ..., Eighth row of the image memory 6 read from g, 20f, 20e ,.
..., written as 8B. When this work is completed, the data array in the image memory 6 is as shown in FIG. 16 (d).
Figure 16 (e). The orientation of the image has been rotated 90 degrees counterclockwise from the initial state (Fig. 16 (a)), and the specified purpose has been achieved.

When each shift register 20a to 20h of the image register 20 is caused to function as a shift register for right shift, the image can be rotated clockwise by 90 ° by the same operation. Also, in the configuration of FIG. 15, serial input / parallel output type is used for each shift register 20a to 20h, but this circuit can also be configured by using parallel input / serial output type.

Next, the overall operation of the image rotation device shown in FIG. 1 will be described. Here, the case where the oblique axis conversion and the enlargement, and the oblique axis conversion and the contraction are simultaneously performed will be described. In the following description, it is assumed that θ is in the range of (−45 ° θ45 °). Now, assuming that the image is rotated by the angle θ, the rotation can be executed by the processing steps of the following four steps. 1st step: oblique axis conversion tan θ and sub-scanning direction expansion Supposing that the line number of the original image is i, the shift amount of the i-th line is S _(i) , and the line number after conversion is j (i, j = 0,1, ...), S _(i) = [tan θ × i + a ] (12) [] is a Gaussian symbol, and finds the largest integer that does not exceed the value in [].

i = [cos θ × j + b] (13) may be converted. Where a and b are 0a and b <1.0
Is the value of.

Rounding, rounding, rounding up, etc. can be determined by the values of a and b. In the conversion equations of the equations (12) and (13), i may be sequentially repeated from 0 to the last line. Image shift part
In 3b, the shiftable bit is up to (W-1), and if it exceeds it, the control unit 3a controls the transfer destination address and transfers to the correct address. That is, S _(i) = A
_(I) × W + B _(i) can be decomposed. A _(i) is the control unit
The number of words to be shifted in 3a, and B _(i) is the number of bits to be shifted in the image shift unit 3b.

Since it takes processing time to solve equations (12) and (13) for each line, prepare two sets of adders with an arbitrary number of bits, one for shift bits and one for enlargement. , For shift bit, tan θ, for expansion Is added every time one line of the original image is advanced (processing of FIGS. 3 and 4). The actual number of additions is [tan θ × 2 ^m ] if the adder has m bits. Is. If a carry occurs in the shift bit adder, the number of shift bits is incremented by one. If a carry occurs in the adder for enlargement, insert the same line. FIG. 19 shows a processing example.

2nd step ... corresponding to the processing of FIG. 5, 90 ° rotating part 3c
Use to rotate the enlarged cross section of the first step by 90 °.

Step 3 ... Corresponds to the processing of FIGS. 6 and 7, the sub-scanning direction reduction angle is cos θ, the oblique axis conversion angle is tan θ, the original image line number i, and the converted line numbers j, i When the shift amount of the row is S _(j) (i, j = 0,1, ...), j = [cos θ × i + c] (14) S _(j) = [tan θ × j + d] ...... ( 15) It is sufficient to repeat the conversion of (0c, d <1.0) from i = 0 to the end. The shift of W bits or more is performed by address control. Also in this case, to perform the processing without solving the equations (14) and (15), prepare two sets of adders as in the first step, one for reduction and one for shift bits. To For reduction, cos θ is added each time the original image advances by one line, and when a carry occurs, one line of the original image at that time is shifted and transferred. At the same time, tan is added to the shift bit adder.
Add θ. If there is a carry at that time, the shift bit is incremented by one. This is repeated below.

Fourth step: Rotate the reduced cross-sectional view made in the third step by -90 ° by using the 90 ° rotating part 3c (corresponding to the process in Fig. 8).

By the above four steps, the image can be rotated by θ.

In the above description, an example in which the rotation processing is performed on the image data read from the input device is shown, but other than this, for example, the present invention can be applied to a character pattern stored in a memory for a character generator. is there. Therefore,
In a device such as CAD, it is possible to easily perform a process of generating a character inclined at a certain angle.

Further, in the embodiment, the original image size is rotated without changing, but if the expressions (5) and (7) are changed as follows, However, c is the enlargement / reduction ratio.

As is clear from the equations (16) and (17), it becomes possible to rotate the original image by θ by enlarging / reducing the original image at an arbitrary magnification. Here, the conversion of equations (16) and (17) is the same as the conversion of equations (5) and (7). It is only necessary to convert the term of cos θ, and the processing procedure and complexity are completely the same as those described above.

〔The invention's effect〕

As described above, according to the image rotation method of the present invention, the image is rotated by performing the conversion of the oblique axis, the enlargement / reduction of the sub-scanning direction, and the 90-degree rotation processing. In addition, rotation to an arbitrary angle can be performed without distortion.

[Brief description of drawings]

1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a block diagram showing an example of an editing apparatus to which the present invention is applied,
3 to 8 are explanatory views showing the processing principle of the present invention, FIG. 9 is a detailed block diagram of the image shift unit 3b according to the present invention, and FIG. 10 is a cyclic shift circuit 12 constituting the image shift unit 3c. FIG. 11 is a block diagram showing the configuration of FIG. 11, FIG. 11 is an explanatory diagram of the shift operation of the cyclic shift circuit, FIG. 12 is a detailed circuit diagram of the data selector 14 that constitutes the image shift unit 3c, and FIG. Detailed circuit diagram of the selector control circuit 15, FIG. 14 is an operation explanatory view showing the operation of the image shift unit 3c, FIG. 15 is a detailed block diagram of the image 90 ° rotation unit 3c, and FIGS. 16 (a) to 16 (e). Is a data array diagram showing the contents stored in the image memories 6a and 6b, FIG. 17 is an operation explanatory diagram of the shift register shown in FIG. 15, and FIGS. 18 (a) and 18 (b) are image data writing to the image register. Explanatory diagram showing the state of
FIG. 19 is an explanatory diagram of simultaneous processing of oblique axis conversion and enlargement according to the present invention. Explanation of symbols 3 ... Image rotation device, 3a ... Control unit, 3b ... Image shift unit, 3c ... Image 90 ° rotation unit, 8 ... Data bus, 9 ... Address bus, 10 ... Register, 11 ...... Data register, 12 …… Circular shift circuit, 13 …… Data latch, 14 …… Data selector, 15 …… Selector control circuit, 16 …… Output bus driver.

Claims (3)

[Claims] 1. An image memory, which stores image data configured by arranging a predetermined plurality of scanning lines, each of which is composed of a predetermined word arranged in the main scanning direction, in the sub-scanning direction, and is stored in the image memory. In the image rotation method of editing and rotating the image data in units of words, the input image data is shifted by a shift amount determined for each scanning line depending on the rotation angle θ and the position in the sub-scanning direction, and the shifted image data Is enlarged at an enlargement rate determined by the rotation angle θ, the enlarged image data is rotated by 90 degrees, the image data rotated by 90 degrees is reduced at a reduction rate determined by the rotation angle θ, and the reduced image data is A shift amount is determined for each scanning line depending on the rotation angle θ and the position in the sub-scanning direction, and the shifted image data is rotated by −90 degrees. Image rotation method. 2. The shift amount is a tangent (tan θ) of a rotation angle θ.
The image rotation method according to claim 1, which is determined by the position in the sub-scanning direction. 3. The enlargement ratio in the enlargement is the reciprocal (1 / cosθ) times the cosine of the rotation angle θ in the sub-scanning direction, and the reduction ratio in the reduction is the cosine of the rotation angle θ in the sub-scanning direction. (Cos θ)
The image rotation method according to claim 1 or 2, wherein the image rotation method is doubled.