JP2001145106A - Device and method for compressing image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve a JPEG based-line compressing method, so that a digital image of high picture quality can efficiently be compressed with as a small resource load as possible. SOLUTION: Respective pixel blocks are subjected to DCT3 and AC coefficients are put in one sequence through zigzag scanning 55. Then the AC coefficient sequence is encoded by Hofmann encoding (57) and linear Hofmann encoding, where zero run-length is not considered and only coefficient values are considered, is carried out here. A linear Hofmann table 11 used here is an extraction of parts of run-length = 0 from a conventional two-dimensional Hofmann table and includes no event for the coefficient value = 0. For the purpose, AC coefficients, having a value '0' are replaced with a value '1', and the Hofmann encoding is carried out. In final bit stream generation 7, frequency information for Hofmann table generation is included in an output bit stream, and the frequency information has contents by which the conventional two-dimensional Hofmann table can be generated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a digital image compression method, and more particularly to an improvement of a JPEG baseline method.

[0002]

2. Description of the Related Art JP as a compression method for digital still images
EG is known, and among a plurality of JPEG compression methods, a baseline method using DCT (Discrete Cosine Transform) is widely used. The procedure of the JPEG baseline method is described in detail in, for example, Japanese Patent No. 2711665.

In this compression method, first, an input bitmap image is divided into blocks of a predetermined size, for example, blocks composed of 8 × 8 pixel values. Next, DCT (discrete cosine transform) is applied to each pixel value block. The DCT generates a new coefficient block including 8 × 8 transform coefficients from each 8 × 8 pixel value block. The first transform coefficient of this coefficient block indicates the magnitude of the DC (direct current) component of the pixel value block, and the other 63 transform coefficients indicate the magnitudes of the AC (alternating current) components having different frequencies. Next, the coefficients in each coefficient block are quantized. Next, a zigzag scan is performed on each coefficient block,
The coefficients of each coefficient block are arranged in a zigzag sequence. Then, two-dimensional Huffman coding (run-length Huffman coding) is performed on the coefficient sequence to reduce the number of bits.

[0004] In the above-mentioned compression system, in order to obtain a high compression ratio, the characteristics of general digital still images are successfully used. That is, when DCT is applied to a general digital still image, in the resulting coefficient block, most of the high-frequency AC transform coefficients have a value of zero or a value close to zero.
Therefore, most of the high-frequency AC transform coefficients become zero by the subsequent quantization. Therefore, the coefficient sequence obtained by the next zigzag scan includes many consecutive columns of zeros. In the subsequent two-dimensional Huffman coding, the number of bits is effectively reduced by performing Huffman coding in consideration of the length (zero run length) of a continuous column of zeros.

That is, in the two-dimensional Huffman coding, a sequence of consecutive zeros appearing in a coefficient sequence and one non-zero coefficient (non-zero coefficient) preceding (or following) it is combined. , As one event to be encoded. Then, according to the principle of Huffman coding, a shorter codeword is assigned to an event having a higher statistical appearance frequency. For example, an event where the zero run length is zero and the bit length of the following non-zero coefficient is one is:
Since the frequency of filing is the highest, the shortest code word is assigned. Also, the event that the zero run length is 1 and the bit length of the subsequent non-zero coefficient is 1 is the next most frequently applied, so that
Assign the next codeword. Similarly, for other events,
A code word according to the appearance frequency is assigned. Thus, the overall number of bits is effectively reduced by two-dimensional Huffman coding that incorporates two dimensions: zero run length and subsequent non-zero coefficients.

[0006] It is also common that all coefficients after a certain position in a coefficient sequence are zero. The first is a block that is completely filled with one color, and all AC coefficients are zero. In such a case,
A code word EOB (end of block) (specifically, coordinates X, Y =
0,0 Huffman codeword). EOB
Means that all subsequent AC coefficients are zero. This EOB further reduces the total number of bits.

In an actual apparatus, a two-dimensional Huffman table 100 as illustrated in FIG. 1 defining events and code words for two-dimensional Huffman coding is prepared on a ROM or a RAM, and the two-dimensional Huffman table is prepared. With reference to 100, each event occurring in the coefficient sequence is converted into a corresponding codeword. As shown in FIG. 1, in the two-dimensional Huffman table 100, each event is represented by a combination (X, Y) of a zero run length X and a bit length Y of a subsequent non-zero coefficient, and each event ( X, Y)
Are described in the frame corresponding to (1) (specific code words are not shown).

[0008]

As described above, the conventional J
The PEG baseline method achieves a high compression effect by using two-dimensional Huffman coding that incorporates a zero-run-length element on the assumption that the sequence of transform coefficients after zigzag scanning contains many consecutive zero columns. I'm trying.

[0009] However, the above assumption is not always satisfied with extremely high quality digital images. For example, in a digital image captured by a digital camera whose image quality has been improved in recent years, the frequency of occurrence of zero before EOB in a coefficient sequence after zigzag scan is extremely low. Therefore, for the output image of the digital camera, the compression effect by adopting the zero run length element in the Huffman compression after DCT is weak. On the contrary, the preparation of the two-dimensional Huffman table increases the resource load of the ROM and the RAM, which is a problem particularly when a compression device is mounted on a device having a strict ROM restriction or an ASIC having a small number of gates.

Accordingly, it is an object of the present invention to provide an image compression method improved from the JPEG baseline method, in which a high compression effect on a high-quality digital image can be obtained with as little resource load as possible.

Another object of the present invention is to provide a JPE image processing apparatus which achieves the above object and which can develop an image using a conventional JPEG baseline image developing apparatus which is already widely used.
An object of the present invention is to provide an image compression method improved from the G baseline method.

[0012]

SUMMARY OF THE INVENTION The improved JP of the present invention
According to the EG baseline compression method, the DC of the digital image
When Huffman coding is performed on the AC coefficient generated by the T coding, a Huffman codeword group assigned to a one-dimensional event group related only to the value of the AC coefficient is used.
Since the Huffman code word group based only on this coefficient value has a smaller number of code words than the conventional two-dimensional Huffman table corresponding to the combination of the coefficient value and the zero run length, a resource burden when expanding the Huffman table is reduced. light. Also, even if the Huffman coding does not incorporate the zero-run-length element, high-quality digital images such as digital camera output images have a low frequency of zeros, so high-quality and high-quality images comparable to conventional methods Compression ratio.

In a preferred embodiment, the Huffman codewords described above do not include codewords assigned for zero AC coefficients. Therefore, when the zero AC coefficient is subjected to Huffman coding, the zero AC coefficient is replaced with another predetermined value (for example, value 1) and converted into a Huffman code word corresponding to the predetermined value. The reason that the Huffman codeword group does not include the Huffman codeword for the coefficient value of zero is that the Huffman codeword group has a run length of zero from the two-dimensional Huffman table used in the conventional JPEG baseline method. This is because only the codeword group in the case is extracted. As described above, by using the group of Huffman code words included in the conventional two-dimensional Huffman table, it is possible to use a conventional JPEG developing device when developing an image. Even if the value of the AC coefficient of zero is replaced with the value of one, it is not perceived by human eyes as image quality deterioration.

In a preferred embodiment, in order to make it possible to use a conventional JPEG decompressor when decompressing an image, furthermore, when generating an output bit stream of the compression result, the two-dimensional The output bit stream contains information for generating a two-dimensional Huffman table that is a Huffman table whose run length is zero and whose part matches the codeword group used in the Huffman coding.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows a flow of an image compression process by an improved JPEG baseline method according to an embodiment of the present invention.

As shown in FIG. 2, this image compression processing
It can be roughly divided into four stages: preprocessing 1, DCT coding 3, entropy coding 5, and bit stream generation 7. In this embodiment, the same method as the conventional JPEG baseline method is used for preprocessing 1, DCT encoding 3, and bit stream generation 7. In the entropy coding 5,
The same method as in the related art is used for the processes 51 and 53 for the DC coefficient. Further, among the processes on the AC coefficients of the entropy coding 5, the zigzag scan 55 uses the same method as the conventional one. Therefore, in this embodiment, a new method is used only for Huffman coding 57 for AC coefficients.

In the following, the conventional method is well known and will not be described in brief. A new method (Huffman coding 5 for AC coefficients) will be described.
7) will be described in detail.

In preprocessing 1, color conversion for converting the color space of the input bitmap image from RGB to YUV (YCbCr), and the image of each of the three color components is performed, for example, in a size of 8 × 8.
Block processing for subdividing into blocks.

In the next DCT coding 3, first, two-dimensional DCT is applied to each block to convert it into, for example, a transform coefficient block of size 8 × 8. Next, each coefficient block is subjected to quantization using a different quantization step for each transform coefficient. The quantization step for each transform coefficient is defined in a quantization table 9 prepared in advance.

In the next entropy coding 5, first, 64 quantized transform coefficients of each 8 × 8 coefficient block are divided into one DC coefficient and 63 AC coefficients. And
For the DC coefficient, after performing differential encoding 51, Huffman encoding 53 is performed. In the differential encoding 51, the DC coefficient of the previous block and the DC coefficient of the current block often take similar values, so that the next Huffman encoding 53 is made more effective.
The value after differential encoding 51 and the Huffman code word assigned to it are defined in the DC Huffman table 11 on the memory.

In the process for the AC component of the entropy coding 5, first, a zigzag scan 55 is performed on the AC coefficient of each coefficient block, and 63 AC signals in each block are performed.
The coefficients are arranged in one sequence in the order of low band first and high band later. Next, Huffman coding 57 is applied to this coefficient sequence. The Huffman coding 57 is a one-dimensional Huffman coding based on only a one-dimensional element called a bit length of a coefficient without adopting a zero-run-length element. The correspondence between the bit length of the coefficient and the Huffman codeword is
It is defined in the AC Huffman table 13 prepared on the memory. This AC Huffman table 13 is similar to the two-dimensional Huffman table 100 illustrated in FIG. 1 used in the conventional JPEG baseline method, in which a portion of zero run length X = 0, that is, only one vertical column at the left end is extracted. , A one-dimensional Huffman table based only on the bit length of the coefficient absolute value.

In the last bit stream generation 7, image information such as the image size and the number of quantization bits, a color space, a quantization table, and a DC Huffman table are added to the sequence of code words generated by the entropy coding 5. 11 generation frequency table and AC Huffman table 13
By adding compression information such as an appearance frequency table for generation,
Output as a bit stream of a predetermined format. Huffman table 1 for AC included in this output bit stream
The contents of the appearance frequency table for generation of the three-dimensional Huffman table 13 shown in FIG. 3 are not limited to the contents of the conventional JP generation table shown in FIG.
What can generate the two-dimensional Huffman table 100 of the EG baseline method, that is, A included in the output bit stream compressed by the conventional JPEG baseline method
The contents are the same as those of the appearance frequency table for generating the Huffman table for C. Therefore, the image compressed in this embodiment can be expanded using a conventional JPEG image expansion apparatus.

Hereinafter, the processing of the Huffman compression 57 for the AC coefficients in the above-described entropy coding 5 will be described in detail. FIG. 3 shows an AC one-dimensional Huffman table 13 used in this processing.

In the Huffman compression 57, the bit length of the absolute value of each coefficient appearing in the coefficient sequence obtained by the zigzag scan 55 is treated as one event to be encoded. In the one-dimensional Huffman table 13 illustrated in FIG. 3, the occurrence frequency of the total 16 types of events in which the bit length Y of the absolute value of the coefficient is each value from 1 to 15 and when the bit length Y is 16 or more is The corresponding code word is assigned. Of these 16 types of events, the event of Y = 1 has the highest frequency of occurrence, so the shortest (for example, 2-bit) code word is assigned, and the event of Y ≧ 16 has the lowest frequency of occurrence, and therefore has the longest. A code word is assigned (in FIG. 3, the code word is not shown). In addition, as in the prior art, an EOB (end of block) codeword meaning "all subsequent AC coefficients are zero" is also prepared (specifically, coordinates X and Y in a two-dimensional Huffman table). = 0, 0 Huffman codeword), which is also a relatively short (eg, 4 bit) codeword.

The one-dimensional Huffman table 13 does not define an event that the coefficient bit length is Y = 0 (that is, the coefficient value is zero). The reason is that only run length = 0 is extracted from the two-dimensional Huffman table 100 of the conventional JPEG baseline method shown in FIG. 1 (in principle, there is no Y = 0 in principle of considering zero run length). This is because this one-dimensional Huffman table 13 is used. As described above, by using the one-dimensional Huffman table 13 having a configuration that matches the conventional two-dimensional Huffman table 100, the image compressed in the present embodiment can be converted into a widely used conventional JPEG image. It is possible to deploy with a deployment device. As will be described later, when a zero not defined in the one-dimensional Huffman table 13 appears before the EOB, a measure is taken by replacing the coefficient value zero with a value “1”.

FIG. 4 shows the overall processing flow of Huffman coding 57 for the AC coefficient sequence for each block.

First, the A obtained by the zigzag scan 55
The end of valid data in the C coefficient sequence is determined (S1). Here, the end of valid data refers to a coefficient in which all coefficients after the coefficient are 0 (that is, EOB is placed immediately after the coefficient). FIG. 5 shows the details of the process for determining the end of the valid data. As shown in FIG. 5, after initially setting a value Index indicating the position (order from the head) of each coefficient in the coefficient sequence to 64 (S21), the Index is reduced by one at a time.
It is checked whether the value of the AC coefficient at the position indicated by x is 0 (S22, S23). That is, whether each AC coefficient is zero is checked in order from the end to the beginning of the coefficient sequence. Then, when the first non-zero AC coefficient is found, the position Index of the AC coefficient is returned to the main processing of FIG. 4 as the end of valid data (S2).
4).

Referring again to FIG. 4, next, each AC coefficient is read in order from the beginning of the coefficient sequence (S2), the absolute value of the read coefficient is determined (S3), and whether or not the absolute value is zero is determined. Is checked (S4). As a result, if the absolute value is zero, the absolute value is changed to "1" (S5). As described above, since the coefficient of zero is not defined in the one-dimensional Huffman table 13, the zero is replaced with the closest value "1". Even if zero is replaced by “1”, there is almost no difference between zero and “1” given to human vision, and especially in a high-quality image such as an output image of a digital camera, zero is included in effective data. Is less frequent, and there is almost no substantial image quality degradation.

Next, the bit length of the absolute value of the AC coefficient is determined (S6). Details of this processing are shown in FIG. As shown in FIG. 6, first, after initializing the bit length Y to Y = 0 (S31), the byte data indicating the absolute value of the coefficient is shifted right by one bit (that is, the absolute value is set to 2). Division), and increments the bit length Y by 1 (S3
3). Next, it is checked whether the absolute value data right-shifted by one bit is zero (S33), and if not, the step S32 is repeated. When the value becomes zero in step S33, the bit length Y obtained by the increment in step S32 is set as the bit length of the absolute value of the coefficient, as shown in FIG.
Is returned to the main process (S34).

Referring again to FIG. 4, next, the Huffman code word corresponding to the bit length Y of the coefficient absolute value returned from the processing of FIG. 6 is read from the one-dimensional Huffman table 13 shown in FIG. S7), the Huffa code word is written into a bit stream to be output (S8). Since the code word data read from the Huffman code table 13 on the memory in step S7 has a byte unit length, in step S8, only the bit length portion of the net code word is extracted from the read bytes. Write to the bitstream.

Next, the sign of the coefficient is checked (S9). If the coefficient is positive, the value of the coefficient itself is written into the bit stream (S11). If the coefficient is negative, the coefficient is subtracted from 1 (step S9). S10), and writes the difference value to the bit stream (S11). At this time, from the bytes of the original coefficient data, only the bit length Y of the net coefficient is extracted and written into the bit stream. Thereby, the number of bits of the coefficient data is reduced. For example, in the case of a coefficient having a bit length of Y = 1, which has the highest frequency of occurrence, it is originally 1 byte, but is shortened to 3 bits obtained by combining the shortest Huffman code word of, for example, 2 bits and the bit length of the net coefficient of 1. Is done.

The above processing of S2 to S11 is repeated until the end of the valid data of the coefficient sequence (S12). Then, it is checked whether the end of the valid data is the 63rd (that is, the end of the coefficient sequence) (S13), and if it is before the 63rd, the EOB codeword is written into the bit stream (S14). On the other hand, if the number is 63, the Huffman coding for the coefficient sequence of the block is terminated, and the process proceeds to the Huffman coding of the next block.

In the embodiment described above, the Huffman coding of the AC coefficients is performed based only on the elements of the coefficient values without considering the elements of the run length, so that only a one-dimensional Huffman table needs to be prepared. As a result, the load on the resources of the RAM or ROM is reduced as compared with the conventional JPEG compression apparatus, and particularly, implementation on an apparatus having a strict memory limit or an ASIC having a limited number of gates is facilitated. For a high-quality image such as an output image of a digital camera in recent years, an image quality comparable to that of the related art can be secured. further,
A one-dimensional Huffman table adapted to the conventional two-dimensional Huffman table is used, and the output bit stream includes frequency information that can generate the conventional two-dimensional Huffman table. Can be performed.

The above embodiment is merely an example for explaining the present invention, and is not intended to limit the present invention to only this embodiment. Therefore, the present invention can be implemented in various modes other than the above-described embodiment. For example, if it is not necessary to perform image expansion using a conventional JPEG expansion apparatus, the frequency information for generating the AC Huffman table to be included in the output bit stream can be simply generated by the AC one-dimensional Huffman table used for compression. The event may be added to the AC one-dimensional Huffman table.

By the way, in this specification, the image compression technique according to the present invention is mainly described, and the image expansion is not particularly described. However, compression and decompression of an image are two sides of the same coin, and decompression can be performed by reversing the compression conversion operation. Therefore, if a person skilled in the art knows the compression conversion operation, the decompression conversion operation is obvious. Therefore, the technical scope of the present invention described in each claim includes not only the compression apparatus and method exactly as described, but also the expansion apparatus and method which are in an integral relationship with the compression apparatus and method.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram showing a two-dimensional Huffman table used for Huffman coding of AC coefficients in a conventional JPEG baseline method.

FIG. 2 is a block diagram showing a configuration of an embodiment of the present invention.

FIG. 3 is an explanatory diagram showing a one-dimensional Huffman table used for Huffman coding of AC coefficients in the embodiment.

FIG. 4 is an exemplary flowchart showing the overall flow of Huffman encoding of AC coefficients in the embodiment.

FIG. 5 is a flowchart of a process of detecting the end of valid data of an AC coefficient sequence.

FIG. 6 is a flowchart of a process for obtaining a bit length of an absolute value of an AC coefficient.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Pre-processing part 3 DCT coding part 5 Entropy coding 7 Bit stream generation part 9 Quantization table 11 One-dimensional Huffman table for DC 13 One-dimensional Huffman table for AC

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5C059 KK08 MA00 MA23 MC01 MC11 MC24 MC33 MC34 MC38 ME02 ME05 PP01 PP15 PP16 SS20 UA02 UA38 UA39 5C078 AA09 BA57 CA01 5J064 AA01 BA08 BA09 BA16 BC01 BC02 BD01

Claims (9)

[Claims] 1. An apparatus for compressing a digital image by a JPEG baseline method, wherein a one-dimensional event group related only to the value of the AC coefficient when the AC coefficient generated by the DCT coding of the digital image is Huffman-coded. The first assigned to each
An image compression apparatus using the Huffman codeword group. 2. The first group of Huffman codewords does not include a Huffman codeword assigned to the event that the AC coefficient is zero, and when the Huffman coding of zero AC coefficient is performed, the zero AC coefficient is used. 2. The image compression apparatus according to claim 1, wherein is replaced with another predetermined value, and is converted into a Huffman code word assigned to an event corresponding to the predetermined value. 3. The image compression apparatus according to claim 2, wherein said another predetermined value is one. 4. When generating an output bit stream containing the result of the Huffman encoding, a two-dimensional event group related to a combination of a non-zero AC coefficient and a run length of a zero AC coefficient is generated. 2. The output bitstream further includes information for generating a second group of Huffman codewords that are assigned and that match the first group of Huffman codewords when the runlength is zero. Image compression device. 5. A method of compressing a digital image by a JPEG baseline method, wherein a one-dimensional event group related only to the value of the AC coefficient when the AC coefficient generated by the DCT coding of the digital image is Huffman-coded. The first assigned to each
An image compression method using a Huffman codeword group. 6. The first group of Huffman codewords does not include a Huffman codeword assigned to the event that the AC coefficient is zero, and when the Huffman coding of zero AC coefficient is performed, the zero AC coefficient is used. 6. The image compression method according to claim 5, wherein is replaced with another predetermined value, and is converted into a Huffman code word assigned to an event corresponding to the predetermined value. 7. The image compression method according to claim 6, wherein said another predetermined value is 1. 8. When generating an output bit stream including the result of the Huffman coding, a two-dimensional event group related to a combination of a non-zero AC coefficient and a run length of a zero AC coefficient is respectively specified. 6. The output bitstream further includes information for generating a second group of Huffman codewords that are assigned and that match the first group of Huffman codewords when the run length is zero. Image compression method. 9. When compressing a digital image by a JPEG baseline method, when performing Huffman coding on an AC coefficient generated by DCT coding of the digital image, a one-dimensional event group related only to the value of the AC coefficient is used. The first assigned to each
A computer-readable recording medium carrying a program for operating a computer so as to use the Huffman codeword group.