JPH08163376A - Picture encoder and picture decoder - Google Patents

Abstract

PURPOSE: To generate a restored picture in the display and print system with the picture size of an objective picture output device independently of the picture size of an original picture. CONSTITUTION: At the time of encoding, an encoded block is inputted from an original picture 50 to a block pattern discriminator 11, and a pattern number correspondence table 31 selected by an encoded block size S is used to express this encoded block by a pattern number, and this pattern number is encoded by an entropy encoder 12 to generate encoded data. At the time of decoding, a block pattern generator 22 uses the pattern number obtained by decoding the encoded data by an entropy decoder 21 and uses a pattern table 41 selected by a designated decoded block size D to generate a decoded block corresponding to the pattern number and successively outputs the generated decoded block to generate a restored picture 70.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is intended to improve the efficiency of storage or communication when an image input from an image input device is stored or communicated and output to an image output device having a different resolution, display and printing method. The present invention relates to an image encoder and an image decoder used for compressing the data amount of an original image.

[0002]

2. Description of the Related Art Since the conventional binary image compression technology has been developed for the purpose of faithfully reproducing the original image and further increasing the compression rate, the restored image should be of the same image size as the original image. It can be decrypted.

In addition, JBIG (Joint Bi-level Image ex
The coding method proposed by ITU as a paerts group) employs a stepwise coding method.

[0004]

In the former prior art,
When the resolution and the display and printing methods of the image input device that created the original image and the image output device that outputs the restored image are different, the target image output device is created by once performing image processing on the decrypted restored image. Need to generate an output image for.

In the latter conventional technique, it is possible to decode a restored image having an image size different from that of the original image, but the magnification selectable as the image size of the restored image is a reduced image of 1 times a power of 2. Therefore, when the resolution of the image output device cannot be obtained at the corresponding reduction ratio or when enlargement is required, it is necessary to add image processing after decoding similarly.

It is an object of the present invention to provide an image coder capable of generating an image size of a target image output device and a restored image of a display and printing method with an image size that does not depend on the image size of an original image. It is in.

[0007]

An image encoder according to the present invention includes first means for inputting image data, each coded block having a coded block size M * N of the input image data, and corresponding thereto. Pattern number correspondence table consisting of pattern numbers corresponding to each other, and the coding blocks are sequentially cut out with the coding block size M * N from the image data input by the first means, and the code is referred to by referring to the pattern number correspondence table. A second means for coding a pattern number corresponding to a coded block as a value of the coded block; and a third means for transmitting the coded block size M * N and a pattern number corresponding to each coded block. Have.

Also, the image decoder of the present invention comprises first means for receiving the coding block size M * N and the coding pattern number sequence from the image coder, and the coding block size M * N. A pattern number conversion table composed of a pattern number and a decoding block in a decoding block size J * K corresponding to each pattern number in the above, and the pattern number conversion corresponding to each pattern number in the encoded pattern number sequence There is provided a second means for obtaining the pattern number of the table, referring to the pattern number conversion table, and decoding the image data using the decoding block corresponding to the pattern number.

The second means of the image encoder has pattern number correspondence tables for the types of combinations of the coding block size M * N and the decoding block size J * K.

The third means of the image decoder has pattern number conversion tables for the types of combinations of the coding block size M * N and the decoding block size J * K.

The second means of the image encoder is arranged to encode the pattern number by entropy coding, and the second means of the image decoder is arranged to decode the pattern number by entropy decoding. It is configured.

A black and white image or a black and white halftone image is handled as an image.

N * N is used as the coding block size and J * J is used as the decoding block size.

[0014]

According to the present invention, the encoding and decoding processes are performed on a block-by-block basis. At the time of encoding, the block is encoded by a value expressed by a pattern number corresponding to the arrangement pattern of pixels in the block, and at the time of decoding it is decoded. The decoded image size is generated by using the value decoded by the coding as a pattern number and the decoded block size and the pixel arrangement pattern in the block different from the coded block size and the pixel arrangement pattern in the block. It is a thing. As the pixel arrangement pattern in the decoding block registered in advance, the decoding block size and the pixel arrangement in the block of the magnification as necessary are selected from the enlarged or reduced arrangement pattern of the pixel arrangement pattern in the code block. By performing the decoding process using the pattern, the original image can be decoded from one piece of encoded data with an image size that can be directly input to the image output device that displays or prints the restored image.

The image may be a monochrome image or a monochrome intermediate image.

[0016]

Embodiments of the present invention will now be described with reference to the drawings.

FIG. 1 is a block diagram of an image encoder / decoder according to an embodiment of the present invention.

The encoder 10 is an encoder for performing resolution-independent encoding, and has a specified encoding block size S.
Accordingly, the S × S blocks are sequentially input from the original image 50, and the block pattern determiner 11 that outputs the pattern number corresponding to the corresponding block and the pattern number output from the block pattern determiner 11 are encoded by entropy encoding. It is composed of an entropy coder 12 for converting. The pattern number correspondence table storage device 30
One or a plurality of pattern number correspondence tables 31 that store correspondences between blocks and pattern numbers used by the block pattern determiner 11 to determine a pattern number from a block are stored according to the coding block size. . The decoder 20 is a decoder that performs resolution-independent decoding, and the entropy decoder 21 that decodes encoded data (pattern / number) by entropy decoding and the entropy decoder 21 The block pattern generator 22 is configured to generate a block corresponding to the converted pattern number with a designated decoded block size D and sequentially output it as a restored image 70. The pattern table storage device 40 stores the pattern number conversion table 41 storing the image pattern of a predetermined block corresponding to the pattern number for the block pattern generator 22 to generate the block from the pattern number, according to the decoding block size D. One or more are accumulated. The coded data storage devices 61 and 62 store the coded data serving as the output of the encoder 10 and the input of the decoder 20, respectively. At the time of encoding, since the original image 50 is encoded for each block by the designated encoding block size S, the block pattern determiner 11 inputs the block from the original image 50 and selects the pattern selected by the encoding block size S. The corresponding block is represented by a pattern number using the number correspondence table 31, and the pattern number is represented by the entropy encoder 1.
Encoded data is created by encoding with 2. At the time of decoding, the pattern number is obtained by decoding the coded data by the entropy decoder 21 and the block pattern generator 22 selects the pattern using the pattern number conversion table 41 selected by the specified decoding block size D. The restored image 70 is created by generating blocks corresponding to the numbers and sequentially outputting the generated blocks. In this encoding and decoding processing, the restored image 70 is specified by specifying the decoding block size D different from the encoding block size S specified in the block pattern determiner 11 at the time of encoding to the block pattern generator 22 at the time of decoding. The decoded image 70 that can be directly input to the image output device for displaying or printing is decoded.

FIG. 2 is a flow chart for explaining the process of the block pattern determiner 11. The block pattern determiner 11 first inputs the image size of the original image 50 and the encoded block size S, which are composed of the width size W and the height size H necessary for the initialization process (steps 101 and 102), and Entropy encoder 1
2 is initialized (step 103). Next, the pointers i and j are initialized (steps 104 and 105), and the blocks Bij are converted from the original image 50 in block units of S × S pixels according to the input coding block size S.
(Step 106), and input each block B
A pattern number is generated for ij (step 10
7), the pattern number value is the entropy encoder 12
(Step 108). In this embodiment, the pattern number correspondence table 31 shown in FIG. 1 is not used in the generation of the pattern number, and the S × S 2 of the block Bij is used.
The value of the bit string obtained by arranging the pixels arranged in one dimension in one dimension is generated as the pattern number value. Next, the pointer j is incremented by +1 (step 109), and j>
It is determined whether or not (W / S) (step 110). j>
If not (W / S), the process returns to step 106 and j>
If (W / S), the pointer i is incremented by +1. Next, it is determined whether i> (H / S) (step 112). If i> (H / S) is not satisfied, the procedure returns to step 105. If i> (H / S), that is, if the coding is completed for all pixels of the original image 50, the termination process of the entropy encoder 12 is performed. (Step 113),
The process of the block pattern determiner 11 ends.

FIG. 3 is a flow chart of the processing of the entropy encoder 12, which is an embodiment using the arithmetic encoding adopted in JBIG as the entropy encoding.

The entropy encoder 12 is shown in FIG.
As shown in, category coding (step 201) and bit coding (step 202) are performed.

The category coding is performed by the flow chart shown in FIG. 3B, and the value of the given pattern number is set to 2
This is a process for encoding the number of bits required to express in a binary number, and the bits required to express the value of the pattern number in a binary number by comparing the pattern number value with the category M as shown in Table 1. The number is obtained, and the code bits, which are the number of bits of on-bit (1) and one off-bit (0), are encoded by arithmetic encoding.

[0023]

[Table 1] First, prior to the processing, the initial value of the category M and the initial value of the template T used in the arithmetic coding are set (step 301). Next, by comparing the pattern number value V and the category value M, the category value M = 2 ⁿ
It is determined whether or not the number of bits required to represent a binary number represented by (n: number of bits) is n or less (step 30).
2). The pattern number value is determined by the determination in step 302
If it cannot be represented by the number of bits represented by the category value M, the bit "1" is arithmetically encoded by the template T (step 303). Next, move the category value M to the left by 1.
Bit shift (M = M << 1 = M * 2), that is, M = 2
^The category value M is updated as ^{n + 1} , 1 is added to the template T, and the template value is also updated (step 304). Then, the process returns to step 302 to compare the updated category value M with the pattern number value V.
If it is determined in step 302 that the pattern number value V can be represented by the number of bits represented by the category value M, the bit "0" is arithmetically encoded by the template T (step 305).

As described above, at the time of decoding, the pattern number value is obtained by counting the number of bits "1" decoded by arithmetic decoding until the bit "0" is decoded by arithmetic decoding. It is possible to know the number of bits required to express V in a binary number. Bit encoding is shown in FIG.
It is performed by the flowchart shown in (c), and the pattern number is encoded by arithmetic encoding with the number of bits obtained by category encoding.

First, a template value T for arithmetic coding used for bit coding is initialized (step 40).
1). Next, the category value M obtained by the category coding (V) (FIG. 3B) is right-shifted by 1 bit (step 402). As a result, the category value is updated to M = M / 2. Next, by making the category value M equal to 0, the pattern number value V advanced in category coding is set to 2
It can be determined that the encoding of the number of bits required to express it in a base number is performed. Therefore, here, the category value M is compared with "0" (step 403). If they are equal, the processing ends. If they are not equal, the bit value Bit to be encoded in the pattern number value V is obtained by taking the logical product (AND) of the category value M and the pattern number value V (step 404). Next, it is determined whether Bit, which is the bit value to be encoded, is "0" or "1" (step 40).
5). Bit to be encoded according to the determination in step 405
Is 0, bit "0" is encoded by arithmetic encoding (step 406). After encoding, the process returns to step 402. Bit to be encoded according to the determination in step 405
If is "1", the bit "1" is encoded by arithmetic encoding (step 407). After encoding, the process returns to step 402.

FIG. 4 is a flowchart of the process of the block pattern generator 22. Block pattern generator 2
2 inputs the image size of the original image, the coding block size S, and the decoding block size D, which consist of the width size W and the height size H required for the initialization process (steps 501 and 502) and entropy. The decoder 21 is initialized (step 503). As a result, the number of horizontal blocks (W / S) and the number of vertical blocks (H / S) of the restored image are obtained, and the entropy decoder 21 is requested to decode the pattern number for each block (step 506). ), Using the obtained pattern number as a key, when the coding block size S and the decoding block size D are different, the pattern number conversion table 41 selected by the decoding block size D is searched to change from D × D. The restored pixel block Bij is generated and sequentially output to generate the restored image 70. When the encoding block size S and the decoding block size D are equal, the value of the bit string obtained by arranging in one dimension of the pattern number in the procedure reverse to the process of the block pattern determiner 11 (FIG. 2) is set to D of the block Bij. The restored image 70 is generated using the blocks Bij obtained by arranging in a two-dimensional array of × D (steps 504 to 511). When the decoding process for all blocks is completed, the entropy decoder 21 is terminated (step 512), and the process of the block pattern generator 22 is terminated.

FIG. 5 is a flowchart of the processing of the entropy decoder 21.

The entropy encoder 12 is shown in FIG.
As shown in, category decoding (step 601) and bit decoding (step 602) are performed.

The category decoding is performed according to the flow chart shown in FIG. 5 (b), decoding is performed bit by bit by arithmetic coding, and an on bit (1) until an off bit (0) is detected as a decoded bit. ), The entropy encoder 12 obtains the number of bits that code the pattern number.

First, prior to the processing, the initial value of the category M and the initial value of the template T used in arithmetic coding are set (step 701). Next, the template T is used for decoding by 1-bit arithmetic coding, and the decoded value is E
(Step 702). Next, when the decoded value E in step 702 is “0”, the pattern number V
Is encoded as a delimiter after encoding "1", which is the number of bits required to express B as a binary number, the category decoding is performed when E becomes "0". It has been finished. Therefore, E is compared with "0" (step 703). If they are equal, the category decoding process ends. If not, step 702
As many as the number "1" is decoded as the decoding value E by M
Is updated by 1 bit (M = M << 1 = M * 2) and the template value T for the next arithmetic decoding is updated (step 704), and the process returns to step 702.

The bit decoding is performed by the flowchart shown in FIG. 5C, and the pattern number value is decoded by decoding the number of bits obtained by the category decoding by arithmetic coding.

First, a template value T for arithmetic coding used during bit decoding is initialized and the pattern number value V is cleared to 0 (step 801). Next, the category value M obtained in the category decoding (FIG. 5B) is right-shifted by 1 bit (step 802). As a result, the category value is updated to M = M / 2. When the category value M becomes “0”, it can be determined that the decoding is performed by the number of bits required to express the pattern number value V in binary. Therefore, here, the category value M is compared with "0" (step 803). If they are equal, the value of V is returned as the pattern number value, and the process ends. If the category value M is not "0", the bit value is decoded by bit arithmetic coding and substituted into E (step 8).
04). Next, it is determined whether E is "0" (step 805). If E is equal to “0”, it is not necessary to set the target bit to the pattern number value V, and the process returns to step 802. If the bit value is "1" in step 805, the category value M represents the corresponding bit position, and therefore the logical sum (OR) of the pattern value V and the category value M is taken to find the corresponding pattern value V. Is set (step 806). FIG. 6 shows an example in which an original image 50 coded with a coding block size S of 2 is subjected to a decoding process with a decoding block size D of 3, and a decoded image 70 that is 1.5 times the original image 50 is restored. It is a figure.

In this example, the original image 50 is coded for each block of 2 × 2 pixels, and as an example, 2 × shown by 51 in FIG.
Focusing on the encoding processing and the decoding processing of the 2-pixel block, the outline of the processing will be described.

For the 2 × 2 pixel block 51 in FIG. 6, the block pattern determiner 11 uses the pattern number correspondence table shown in FIG.
4 is obtained. The pattern number 14 is entropy-encoded by the entropy encoder 12 in the flowchart of FIG. 3 to obtain an entropy code.

At the time of decoding, the pattern number 14 is obtained by entropy decoding the encoded entropy code by the entropy decoder 21 in the flowchart of FIG. The block pattern generator 22 is shown in FIG.
Decoding is performed by using the pattern number conversion table shown in (1) to obtain the decoding block 71 composed of 3 × 3 pixels corresponding to the pattern number 14.

By subjecting each block to such processing, the original image 50 of FIG. 6 is restored to the decoded image 70.

As shown above, FIG. 6 shows an example in which the decoded image 70 of 1.5 times different from the image size of the original image 50 is restored by using the pattern number conversion table shown in the figure. By using the pattern number conversion table of, the decoded image at the magnification corresponding to the pattern number conversion table can be restored.

[0038]

As described above, according to the present invention, the block size at the time of coding and the block size at the time of decoding can be arbitrarily selected by using the coding on a block-by-block basis. By generating a restored image with an image size that is a ratio of the block size, by specifying the decoding block size, the image size of the target image output device and the restored image of the display and printing method depend on the image size of the original image. There is an effect that it can be generated with any image size.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an image encoder / decoder according to an embodiment of the present invention.

FIG. 2 is a flowchart of processing of a block pattern determiner 11 in the embodiment of FIG.

3 is an entropy encoder 1 in the embodiment of FIG.
It is a flowchart of the process of 2.

4 is an entropy decoder 2 in the embodiment of FIG.
It is a flowchart of the process of 1.

5 is a flowchart of a process of a block pattern generator 22 in the embodiment of FIG.

FIG. 6 is an explanatory diagram illustrating an operation example of the embodiment of FIG.

[Explanation of symbols]

 10 Encoder 11 Block Pattern Determinator 12 Entropy Encoder 20 Decoder 21 Entropy Decoder 22 Block Pattern Generator 30 Pattern Number Correspondence Table Storage Device 31 Pattern Number Correspondence Table 40 Pattern Number Conversion Table Storage Device 41 Pattern Number Conversion table 50 Original image 51 2 × 2 pixel block 61, 62 Encoded data storage device 70 Restored image 71 3 × 3 pixel decoding block S Encoding block size D Decoding block size

Claims (12)

[Claims] 1. A first means for inputting image data, a pattern number correspondence table comprising each coding block of a coding block size M * N and a pattern number corresponding to the coding block, the first means. A coded block is sequentially cut out with a coded block size M * N from the image data input by, and a pattern number corresponding to the coded block is coded as a value of the coded block with reference to the pattern number corresponding table. An image encoder having a second means for performing the above, and a third means for transmitting the coded block size M * N and a pattern number corresponding to each coded block. 2. A first means for receiving a coded block size M * N and a coded pattern number sequence from the image encoder according to claim 1, and each pattern number in the coded block size M * N. Has a pattern number conversion table composed of a pattern number in the decoding block size J * K and a decoding block corresponding to, and the pattern number of the pattern number conversion table corresponding to each pattern number of the encoded pattern number sequence An image decoder having a second means for obtaining the image data, referring to the pattern number conversion table, and using the decoding block corresponding to the pattern number. 3. The image encoder according to claim 1, wherein the second means has the pattern number correspondence tables for the types of combinations of the coding block size M * N and the decoding block size J * K. . 4. The image decoder according to claim 2, wherein the second means has the pattern number conversion tables for the types of combinations of the coding block size M * N and the decoding block size J * K. . 5. The image encoder according to claim 1, wherein the second means is configured to encode the pattern number by entropy coding. 6. The image decoder according to claim 2, wherein the second means is configured to decode the pattern number by entropy decoding. 7. The image encoder according to claim 1, wherein a black and white image is treated as the image. 8. The image decoder according to any one of claims 2.4.6, wherein a black and white image is treated as the image. 9. The image encoder according to claim 1, wherein a black and white halftone image is treated as the image. 10. The image decoder according to claim 2, wherein a black and white halftone image is used as the image. 11. M of the coding block size M * N
The image encoder according to claim 1, wherein N and N are equal. 12. The J of the decoding block size J * K
7. The image decoder according to claim 2, wherein K and K are equal.