JP3001758B2 - Huffman encoding device and Huffman encoding method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a Huffman coding apparatus and a Huffman coding method for coding data subjected to discrete cosine transform into a Huffman code.

[0002]

2. Description of the Related Art Image data contains a very large amount of information. Therefore, it is not practical to process the image data as it is in terms of memory capacity and communication speed. Therefore, image data compression technology becomes important.

[0003] J is one of the international standards for image data compression.
PEG (Joint Photographic Ex)
part group). DCT (Discrete Cosine Transform) for lossy coding and DP in two-dimensional space
A lossless encoding system that performs CM (Differential PCM) is employed. Hereinafter, DCT image data compression will be described.

FIG. 5 is a block diagram showing a basic configuration of a system for executing image data compression of the DCT system.

[0005] The DCT converter 10 converts the input original image data into a discrete cosine transform (hereinafter referred to as a DCT transform).
Then, a DCT coefficient is output. In this DCT transform, first, as shown in FIG. 6, image data is divided into a plurality of 8 × 8 pixel blocks. As shown in FIG. 7, one 8 × 8
In the pixel block, 64 pieces of pixel data P _XY (X, Y
= 0,..., 7). When two-dimensional DCT transform is performed on each of the divided 8 × 8 pixel blocks, 64 D
The CT coefficient S _UV (U, V = 0,..., 7) is obtained.

[0006] The DCT coefficient _S00 is called a DC coefficient, and the remaining 63 DCT coefficients are called AC coefficients. As shown in FIG. 7, the higher the frequency, the higher the horizontal frequency component of the DCT-transformed block from left to right, and the higher the vertical frequency component of the DCT-transformed block.

[0007] The quantizer 20 of FIG.
, Quantizes the DCT coefficients output from
Output CT coefficients. The image quality and the amount of encoded information are controlled by the quantization. FIG. 8 shows an example of the DCT coefficient output from the quantizer 20. In FIG. 8, "A",
“B”, “C”, “D”, “E”, and “F” represent values other than “0”.

[0008] The encoder 30 in FIG. 5 encodes the DCT coefficients output from the quantizer 20 into Huffman codes, and outputs encoded data. In the coding of the DC coefficient, a difference value between the DC coefficient of the immediately preceding block and the DC coefficient of the current block is obtained, and a Huffman code is assigned to the difference value.

In the coding of AC coefficients, as shown in FIG. 9, first, AC coefficients are arranged one-dimensionally by zigzag scanning. This one-dimensionally arranged AC coefficient is
Encoding is performed using a run length indicating the length of successive “0” coefficients (ineffective coefficients) and values of coefficients (effective coefficients) other than “0”. Effective coefficients are grouped, and a group number is assigned to each effective coefficient. In the coding of an AC code, a Huffman code is assigned to a combination of a run length and a group number. As described above, the original image data is encoded into encoded data.

[0010]

In the DCT-transformed data (DCT coefficients), the low-frequency component represents the color or brightness uniformly existing in the block, and the high-frequency component represents the degree of change in the color or brightness. It represents. Human vision is generally insensitive to high frequency components. Therefore, in the Huffman encoding process, a method of truncating DCT coefficients of a certain frequency or higher has been used in order to improve the compression ratio of encoded data to original image data.

However, when the above method is used, there is a problem that block distortion or the like in which the color or brightness changes abruptly between blocks according to the improvement of the compression ratio, and the image quality is significantly deteriorated. Such deterioration in image quality is particularly remarkable in an image having many high frequency components such as a character image and a computer graphics image.

SUMMARY OF THE INVENTION It is therefore an object of the present invention to provide a Huffman coding apparatus and a Huffman coding method capable of increasing a compression ratio while maintaining image quality.

[0013]

Means for Solving the Problems (1) First invention A Huffman coding apparatus according to a first invention is a Huffman coding apparatus for coding discrete cosine transformed data into Huffman codes in block units. And an encoding unit, a counting unit, a storage unit, and a command unit.

The encoding means includes a discrete cosine transform for each
Encode the block data into Huffman code. The counting means counts the number of consecutive zeros in the data of each block subjected to the discrete cosine transform. The storage means stores in advance the number of consecutive zeros that serves as a reference for data truncation as a reference value. The command means instructs the coding means to end the coding process in the block when the number of consecutive zeros obtained by the counting means becomes equal to or larger than the reference value stored in the storage means.

(2) Second invention A Huffman coding method according to a second invention is a Huffman coding method for coding discrete cosine transformed data into a Huffman code in block units. including. The data of each block subjected to the discrete cosine transform is encoded into a Huffman code, and the number of consecutive 0s in the data of each block subjected to the discrete cosine transform is counted. The number of consecutive 0s, which is a reference for data truncation, is set in advance as a reference value. When the counted number of consecutive 0s is equal to or greater than the set reference value ,
The encoding process is terminated.

[0016]

The present inventor has analyzed a large number of discrete cosine transformed data. As a result, it has been found that when 0s continue for a certain number or more, even if data of higher frequency components is cut off, there is almost no effect on image quality. Obtained. This means that in the data subjected to the discrete cosine transform, if 0 continues more than a certain number, there is almost no data that affects the image quality on the higher frequency side. Therefore, the present inventor has created the present invention paying attention to such a relationship between the number of continuous 0s and the high-frequency component.

In the Huffman coding apparatus according to the first invention and the Huffman coding method according to the second invention, the number of consecutive zeros in the data subjected to the discrete cosine transform is counted, and the number of consecutive zeros is determined in advance. When the value becomes equal to or larger than the set reference value, the Huffman encoding process is terminated.

As a result, when the number of consecutive zeros exceeds the reference value, the data of the high-frequency component is not processed thereafter, and the number of Huffman codes to be encoded decreases. In this case, since the data on the high frequency side hardly affects the image quality, block distortion does not occur, and the compression ratio can be improved while maintaining the image quality.

Also, since the amount of high-frequency component data that is not processed for each block changes according to the characteristics of the image,
Uniform image quality can be obtained. Further, by changing the reference value according to the type of the image, an appropriate image quality and compression ratio according to the type of the image can be obtained.

[0020]

FIG. 1 is a block diagram showing the configuration of a Huffman coding apparatus according to this embodiment. This Huffman encoding device includes a zero data detecting unit 1, a zero counting unit 2, an encoding unit 3,
It includes a zero run length setting section 4 and a block completion detecting section 5. Each of these units can be configured only with hardware or only software, or can be configured with both hardware and software.

The quantized DCT coefficients shown in FIG. 8, for example, are sequentially supplied to the zero data detecting section 1 by zigzag scanning as input data. Zero data detector 1
Detects "0" data (invalid coefficient) from the input data, supplies the detected "0" data to the zero counting unit 2, and outputs data (effective coefficient) other than "0" to the encoding unit 3. Give to.

The zero counting section 2 counts the number of "0" data supplied from the zero data detecting section 1 and supplies the count result to the encoding section 3 and the block completion detecting section 5 as a zero run length. The encoding unit 3 includes a zero data detection unit 1
The Huffman encoding process is performed in the same manner as in the related art based on the effective coefficient given from the above and the zero run length given from the zero counting unit 2 to output encoded data.

On the other hand, in the zero run length setting section 4, a zero run length which is a reference for data truncation is set in advance as a reference value. The image quality improves as the reference value increases, but the compression ratio decreases. The block completion detecting unit 5 compares the zero run length given from the zero counting unit 2 with the reference value set in the zero run length setting unit 4, and sends a block completion signal to the encoding unit 3 when the zero run length becomes equal to or longer than the reference value. The encoding unit 3 ends the Huffman encoding process for the block currently being processed in response to the block completion signal.

Next, the effectiveness of the system of the above embodiment will be described with reference to three types of DCT coefficient blocks shown in FIGS.

FIGS. 2A and 2B show examples of blocks in which no data other than "0" exists in the high-frequency region.
In this type of block, a long zero run length appears in the intermediate frequency region, and no data other than "0" exists in the high frequency region. Therefore, the high frequency component does not affect the image quality.

FIGS. 3A and 3B show examples of blocks in which data other than "0" is distributed from the low frequency region to the high frequency region. In this type of block, a long zero run length does not exist in the intermediate frequency region, and many data other than “0” also exist in the high frequency region. Therefore, the high-frequency component affects the image quality, and if the high-frequency component is cut off, the image quality deteriorates.

FIGS. 4A and 4B show examples of blocks in which no data other than "0" exists in the intermediate frequency region.
In this type of block, a long zero run length appears in the intermediate frequency region, and data other than "0" slightly exists in the high frequency region. In this case, the high frequency component hardly affects the image quality.

The above example also shows that the effect of the high frequency component on the image quality depends on the zero run length.

In the block of the type shown in FIG. 2, the high-frequency component is cut off by using the method of this embodiment or the conventional method, and the data amount is reduced. In this case, since there is no data other than "0" in the high frequency region, the image quality is not affected even if the high frequency component is cut off.

When a conventional method is used for a block of the type shown in FIG. 3, high-frequency components are forcibly cut off, resulting in deterioration of image quality. On the other hand, when the method of this embodiment is used, the zero-run length is short in the intermediate frequency region, so that the high-frequency component is not cut off. Therefore, although the data amount does not decrease, the image quality is maintained.

When a conventional method is used for a block of the type shown in FIG. 4, high-frequency components are cut off, and the amount of data is reduced. Further, even when the method of the present embodiment is used, the zero-run length becomes longer in the intermediate frequency region, so that the high-frequency component is truncated and the data amount is reduced. In this case, since the high-frequency component hardly affects the image quality, the image quality is maintained even if the high-frequency component is cut off.

As described above, the method of this embodiment is very effective for the block of the type shown in FIG. 3, and maintains high image quality as compared with the conventional method.

The reference value set in the zero run length setting section 4 can be arbitrarily selected according to the type of image. For example, in the case of a natural image, high image quality can be obtained by setting a large reference value. On the other hand, in the case of a character image or a computer graphics image having a large high-frequency component, the image quality is maintained even if the reference value is set small. In this case, the compression ratio increases and the processing speed improves.

As described above, the Huffman encoding apparatus of the above embodiment is effective for character images and computer graphics images having large high frequency components. Therefore, by selectively changing the reference value set in the zero run length setting unit 4 according to the type of image, an appropriate image quality and compression ratio according to the type of image can be obtained.

It should be noted that the present invention is not limited to the JPEG system, but includes, for example, MPEG (Moving Picture Image Coding Ex
The present invention is also applicable to other systems using Huffman coding such as a pert group system.

[0036]

As described above, according to the present invention, the Huffman coding process is terminated when the number of consecutive 0s becomes equal to or more than the set reference value, so that the image quality is maintained.
It is possible to improve the compression ratio of Huffman encoded data to original image data. Further, since the amount of high-frequency component data that is not processed for each block changes according to the characteristics of the image, uniform image quality can be obtained. Furthermore, by selecting a reference value according to the type of image, an appropriate image quality and compression ratio according to the type of image can be obtained.

Therefore, in both the Huffman encoding process and the Huffman decoding process, the processing data is reduced, and the processing speed can be increased while maintaining the image quality.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a Huffman encoding device according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating an example of a block of DCT coefficients in which data other than “0” does not exist in a high frequency region.

FIG. 3 is a diagram illustrating an example of a block of DCT coefficients in which data other than “0” is distributed from a low frequency region to a high frequency region.

FIG. 4 is a diagram illustrating an example of a block of DCT coefficients in which data other than “0” does not exist in an intermediate frequency region.

FIG. 5 is a block diagram showing an image data compression system based on the DCT method.

FIG. 6 is a diagram showing blocking of image data.

FIG. 7 is a diagram illustrating an 8 × 8 pixel block and a block subjected to DCT conversion.

FIG. 8 is a diagram illustrating an example of quantized DCT coefficients.

FIG. 9 is a diagram for explaining zigzag scanning.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Zero data detection part 2 Zero counting part 3 Encoding part 4 Zero run length setting part 5 Block completion detection part The same code in each figure shows the same or equivalent part.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N 7/30 G06T 9/00 H03M 7/40 H04N 1/41

Claims (2)

(57) [Claims] 1. A block of discrete cosine transformed data
A Huffman encoder for encoding the Huffman code using the clock unit, the data of the discrete cosine transformed blocks and encoding means for encoding the Huffman code, the discrete cosine transformed in blocks Counting means for counting the number of consecutive zeros in the data; storage means for previously storing the number of consecutive zeros as a reference for data truncation as a reference value; and the number of consecutive zeros obtained by the counting means A Huffman encoding apparatus comprising: a command unit that instructs the encoding unit to end the encoding process in the block when the reference value stored in the storage unit is equal to or more than the reference value. 2. A method for dividing discrete cosine transformed data into blocks.
A Huffman coding method for encoding the Huffman code using the clock unit continuously, while encoding the data of the discrete cosine transformed blocks to the Huffman code, the data of the discrete cosine transformed blocks The number of consecutive zeros to be counted as a reference value is set in advance as a reference value, and when the counted number of consecutive zeros becomes equal to or greater than the set reference value, A Huffman encoding method for ending the encoding process in the block .