JPH04326254A - Image data compressing and encoding device and encoding method - Google Patents

Abstract

PURPOSE:To present the image data encoding device and encoding method to encode image data within an objective code amount in fixed processing time according to a simple method while guaranteeing compatibility with an international standard plan system and maintaining picture quality. CONSTITUTION:The image data encoding device to divide the image data into plural blocks, to execute orthogonal transformation for each block, to quantize the transformed output and to encode it with a variable length executes a first encoding processing with temporary quantizing width at a variable length encoding part 4, optimizes the quantizing width and predicts the allocated code amount for each block based on this first processing by a quantizing width prediction part 9. According to the optimum quantizing width, a specified variable length code is allocated to an EOB(End of Block) by a code amount allocation part 10 and the image data are encoded by executing second encoding while interrupting the encoding so as not to exceed the allocated code amount by using the two kinds of margin code amounts for the final block allocated by an objective code amount in advance and for the middle block.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding apparatus and method for highly compression coding image data.

0002

[Prior Art] Generally, CCD (Charge Co.
When storing an image signal captured by a solid-state imaging device, such as an uploaded device, as digital data in a storage device such as a memory card, magnetic disk, or magnetic tape, the amount of data is enormous, so it is difficult to In order to record data within the available storage capacity, it is necessary to perform some kind of highly efficient compression on the obtained image signal data.

Similarly, when storing digital data on a storage medium such as a memory card, magnetic disk, or magnetic tape using a digital electronic still camera, the data is recorded on one memory card, magnetic disk, or one roll of magnetic tape. The number of images that can be created is specified. Moreover, recording of this specified number of images must be guaranteed, and the time required to record and reproduce image data is short.
Moreover, it needs to be constant.

Furthermore, when recording moving images on a digital VTR (video tape recorder), digital moving image film, etc., it is necessary to be able to record various amounts of frames without being affected by the amount of image data per frame.

That is, whether it is a still image or a moving image, it is necessary to be able to reliably record the required number of frames, and the time required for data recording and reproduction processing is short and constant. There is a need.

An encoding method that combines orthogonal transform encoding and variable length encoding is widely known as a highly efficient image data compression method used in such data recording and reproducing processing. . A typical example is a method being considered in the international standardization of still image coding, and an overview of this method will be explained below.

First, image data is divided into blocks of a predetermined size, and each divided block is subjected to orthogonal transformation.
A two-dimensional DCT (discrete cosine transform) is performed. Next, linear quantization is performed according to each frequency component, and Huffman encoding is performed as variable length encoding on this quantized value. At this time, regarding the DC component, the difference value between it and the DC component of the neighboring block is Huffman encoded. In addition, AC components are scanned from low frequency components to high frequency components called a zigzag scan, and two-dimensional Huffman is calculated from the number of consecutive invalid components (value "0") and the values of the following valid components. Perform encoding.

The above is the basic part of this method. However, in this basic part alone, the amount of code is not constant for each image because Huffman coding, which is variable length coding, is used. Therefore, the Newton-Raphson iteration (Newto
A method using n Raphson Iteration) has been proposed.

However, even if this method is used,
Because the number of times the basic encoding passes are repeated differs depending on the image, not only is the processing time unstable;
Generally, the disadvantage is that the processing time is long.

[0010] As a method for solving these problems, Japanese Patent Application No. 01-283761 and Japanese Patent Application No. 02-02 filed by the present applicant have been proposed.
proposed the method described in No. 137222. These methods are compression methods that combine orthogonal transformation and variable length coding, and in order to control the amount of generated code, an image signal stored in an image memory and sampled is divided into blocks.

After performing orthogonal transformation for each divided block, the outputted transform output is quantized with a provisional quantization width, and then the quantized output is variable-length encoded, and The amount of generated codes for the image and the total amount of generated codes for the entire image are calculated. Next, a new quantization width is predicted from the provisional quantization width, the total generated code amount, and the target total code amount. Further, the allocated code amount for each block is calculated from the generated code amount for each block, the total generated code amount, and the target total code amount.

Then, using the new quantization width, the image signal in the image memory is again divided into blocks, orthogonally transformed, quantized,
When variable length encoding is performed and the amount of code generated for each block exceeds the allocated amount of code for each block, the variable length encoding is stopped midway and processing proceeds to the next block. Thereby, the code amount is controlled so that the total generated code amount for the entire image does not exceed the target total code amount.

In addition, instead of calculating and storing the generated code amount for each block,
There is a method described in No. 3924 that allocates a fixed amount of code to each block.

[0014]

[Problem to be Solved by the Invention] However, if the above-mentioned method is adopted, very good results can be obtained, but in order to effectively utilize the compressed data obtained by this method for a wider range of applications, it is necessary to It is necessary to ensure compatibility with the above-mentioned international standard draft method.

In this international standard draft system, when the compressed data is considered in units of bytes, it is decided to insert a hexadecimal number "FF" as a prefix to recognize a special code.

[0016] However, if the code itself is divided into bytes, there is a possibility that "FF" will be lined up, so it is necessary to distinguish it from the prefix of a special code even if it is not decoded. When it is delimited along with "FF", it is decided that a hexadecimal number "00" is inserted immediately after it (byte stuffing).

Furthermore, when a Huffman code is used as a variable length code, a bit "1" is embedded at the position where the prefix of the special code is inserted so that the code length is aligned in bytes (bit stuffing). st
uffing).

However, it is not known whether or not byte stuffing will occur until the encoding is actually performed and the data is divided into bytes, which makes it difficult to predict the amount of code that will occur and to allocate the amount of code appropriately. Instead, when the generated code amount of each block exceeds the allocated code amount of each block as in the above-mentioned code amount control method (Japanese Patent Application No. 01-283761, Patent Application No. 02-137222), Byte stuffing occurs because variable length encoding of a block is stopped and an EOB (end of block) code is added to indicate the end of the block.
For this reason, the allocated code amount may be greatly exceeded.

[0019] Furthermore, the above-mentioned code amount control method (patent application
2-293724), if the allocated code amount is insufficient to encode even the DC component, only the DC component and EOB code are encoded, and the insufficient code amount is used from the allocated code amount of the next block. Subtracted. However, unless the compression ratio is extremely high, this phenomenon rarely occurs. That is, since this occurs in a situation where the image quality of the entire image is already degraded, the effect on subsequent blocks whose code amount is reduced is not so noticeable.

However, if the shortage of code amount due to byte stuffing is handled in the same way, the amount of usable code will fluctuate greatly, and there is a risk that image quality deterioration will be noticeable. If byte stuffing occurs frequently during encoding, the target total code amount may be exceeded. Depending on the application, these methods may not be applicable.

[0021] Therefore, the present invention aims to encode an image to within a target code amount within a certain processing time using a simple method that maintains image quality while ensuring compatibility with the international standard system. An object of the present invention is to provide a data encoding device and encoding method.

[0022]

[Means for Solving the Problems] In order to achieve the above object, the present invention divides image data into a plurality of blocks,
In an image data encoding device that performs orthogonal transformation for each block, quantizes the transformed output by a quantization means, and variable-length encodes it by a variable-length encoding means, each of the variable-length encoded transform coefficients is A code amount calculation means that calculates the code amount of each block and the code amount of the entire image, and an optimal quantization width based on the code amount of the entire image calculated by the code amount calculation means and the target code amount. a quantization width prediction means for predicting the quantization width, a code amount for each block calculated by the code amount calculation means, and a code amount allocated to each block from the quantization width predicted by the quantization width prediction means; code amount allocation means for allocating a specific code to an end of block (EOB) code indicating the end of the block, and variable length code table generation means for generating a variable length code table; an excess/deficiency code amount calculating means for calculating the accumulation of excess/deficiency of code amount caused by encoding of each block before the added block; From the allocated code amount and the margin code amount allocated from the preset target code amount,
an allocated code amount determining means for determining an allocated code amount for the block; a coding aborting means for terminating the variable length encoding so as not to exceed the allocated code amount determined by the allocated code amount determining means; code output means for multiplexing and outputting the code output from the long encoding means and other additional information; and code output means for causing the quantization means to perform quantization using a preset provisional quantization width. and a control means for controlling the first encoding process to be executed and the second encoding process to be executed using the optimum quantization width and allocated code amount obtained as a result. equipment is provided.

[0023] Also, an image data code in which image data is divided into a plurality of blocks, orthogonal transformation is performed for each block, the transformed output is quantized by a quantization means, and then variable length encoded by a variable length encoding means. The encoding method includes a first step of performing encoding processing on image data and checking the amount of code for each block and the amount of code for the entire image;
The second step is to predict the optimal quantization width and the allocated code amount for each block from the information obtained in the second step, and the third step is to perform quantization again using the quantization width predicted in the second step. step, the allocated code amount predicted in the second step, and the excess or deficiency of the code amount that occurred before that block,
The fourth step is to determine the allocated code amount for the block based on the margin code amount allocated in advance from the target code amount, and variable length encoding is performed to fit within the allocated code amount determined in this fourth step. A method for encoding image data is provided, comprising a fifth step of:

[0024]

[Operation] According to the image data encoding device and encoding method configured as described above, the amount of generated codes for each block and the amount of generated codes for the entire image are calculated, and the optimal quantization width is predicted from these. and also assigns a code amount to each block (first
After quantizing the orthogonally transformed transform coefficients using a new quantization width, the amount of codes allocated to each block by the first encoding and the amount of codes up to the previous block are calculated. The final amount of code allocated to the block is determined based on the surplus or deficiency of the amount of code and the remaining amount of margin allocated in advance from the target amount of code, and while performing byte stuffing if necessary, the amount of code allocated before exceeding the allocated amount of code is determined. , the encoding of that block is discontinued, and variable length encoding is performed (second encoding). Two of these first encoding and second encoding
Pass processing allows encoding to be performed within a certain processing time, and the amount of code is controlled using the allocated code amount for each block, excess or deficiency of the allocated code amount up to that block, and the margin code amount. truncation is performed on high frequency components that do not affect image quality deterioration.

Therefore, according to the present invention, while supporting byte stuffing in order to ensure compatibility with the international standard draft method, it is possible to keep the code amount within the target code amount within a certain processing time with a simple configuration. Thus, image data can be encoded with almost no deterioration in image quality.

[0026]

Embodiments Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram showing the configuration of an image data compression encoding device as an embodiment of the present invention.

[0027] In this image data compression encoding device,
The orthogonal transformer 1 divides the image data input from the input terminal A into blocks, performs orthogonal transform on each block, and outputs the image data as transform coefficients to the quantizer 2. Then, the quantization unit 2 quantizes the transform coefficients according to a quantization table that defines a quantization width for each frequency component, and outputs the quantization coefficients to the variable length code table generation unit 3 and the variable length encoding unit 4. .

Next, the variable length code table generation section 3
A Huffman code table is generated from the frequency of occurrence of the quantized transform coefficient values, and is output to the variable length encoder 4 and the code output unit 7, which will be described later.

In the variable length encoding section 4, after variable length encoding the transform coefficients quantized by the quantization section 2, the bit stuffing section 5 performs variable length encoding at the position where the above-mentioned special code is inserted. A bit "1" is added so that the code lengths are aligned in byte units.

Furthermore, when the code length is output from the bit stuff section 5 to the byte stuff section 6 and the code length is divided into bytes, if a hexadecimal number "FF" exists in the code, Byte “00” is inserted and output to the code output unit 7 and code amount calculation unit 8. The code output unit 7 outputs the encoded image data as described above, the quantization table, the Huffman code table,
It is multiplexed with additional information such as a special code and output to output terminal B.

[0031]The code amount calculation unit 8 calculates the code amount for each block, each component, the entire image, etc.
The calculated code amount is output to the quantization width prediction unit 9. The quantization width prediction unit 9 predicts the optimum quantization width from the calculated code amount from the code amount calculation unit 8 and a preset target code amount, and outputs it to the quantization unit 2.

The code amount allocation unit 10 uses the code amount calculated by the code amount calculation unit 8, the target code amount, and the coefficients used for optimizing the quantization width by the quantization width prediction unit 9. Then, the amount of code is assigned to each block. The encoding abort control unit 11 controls the operation of the variable length encoder 4 so that the encoding is aborted without the generated code amount of each block exceeding the allocated code amount. Next, the encoding method of the image data compression encoding apparatus configured as described above will be explained with reference to FIG.

First, image data (a
part) is divided into blocks of 8×8 samples (hereinafter referred to as block image data), and the orthogonal transform unit 1 performs two-dimensional orthogonal transform using discrete cosine transform (DCT) for each block. The orthogonally transformed block image data (transformation coefficients) are divided into corresponding frequency component positions on an 8 x 8 matrix (the origin position of the matrix is the DC component, the others are AC components, and the frequency increases as the distance from the origin position increases) ) (part b), and this is input to the quantization unit 2.

The quantization section 2 performs the first quantization on this block image data (transform coefficients) (section c).
. In this device, a quantization width is used, which is obtained by multiplying a quantization matrix storing a quantization width for each frequency component set in advance by a provisional quantization coefficient α given by a quantization width prediction unit 9. (Part i).

[0035] Then, the quantized block image data (
The transform coefficients) are input to the variable length encoder 4 and scanned from low frequency components to high frequency components in the order shown in FIG.

The first data in the scanning order shown in FIG. 3 is a DC component, and the difference value diff-
DC is Huffman encoded (d1 part, e1 part). For the AC component, search for conversion coefficients in order from the 2nd to the 64th in the scanning order in FIG. 3 (zigzag scan).
When a valid coefficient (a coefficient that is not "0") appears, a two-dimensional Huffman is calculated using the number of consecutive invalid coefficients (coefficients that are "0") that existed immediately before it (zero run) and the value of the valid coefficient. Encoding is performed (d2 part, e2 part).

Furthermore, if invalid coefficients continue after a certain coefficient up to the 64th coefficient, the variable length encoder 4 adds an EOB code indicating the end of the block. Then, for each code, the code length is calculated by the code amount calculation unit 8.
Output to (g1 part).

[0038] The code amount calculation unit 8 receives input Y, Cr, C.
In order to calculate the code amount for the entire image of each component of b, the code amount for each block is calculated and the code amount is integrated (part g2), and the code amount data for each block is allocated to the code amount. output to section 10. When the Huffman encoding process has been completed for all blocks of one image in this way, the code amount calculation unit 8 outputs the code amount data for the entire image to the quantization width prediction unit 9.

The quantization width prediction unit 9 sets the optimum quantization width coefficient α to the quantization width coefficient used in the first quantization from the code amount data of the entire image and the target total code amount. Prediction is made based on this (h1 part).

[0040] The code amount allocation unit 10 calculates the usable code amount for each block from the first and optimized quantization width coefficient α and the generated code amount by the first encoding for each block. Calculate (h2 part). This completes the first encoding (statistical processing) for predicting the allocated code amount and optimal quantization width for each block. Next, second encoding (encoding processing) is performed to obtain a final encoded output optimized to fit within the target total code amount.

The quantization unit 2 uses the optimal quantization coefficient α (h1 part) and quantization matrix calculated by the quantization width prediction unit 9 in the first encoding for the block image data (transform coefficients) (b part). (part i) is used to perform quantization again (part c).

Next, the quantized block image data (transform coefficients) is input to the variable length encoding section 4. As in the statistical processing, first the difference value diff-DC of the DC component is Huffman encoded (d1 part, e1 part), then the data of the AC component is sequentially extracted by zigzag scan and two-dimensional Huffman encoding is performed ( d2 part, e2 part).

However, for each block processed, the code amount allocation unit 10 calculates the allocated code amount (h2 part) of each block calculated by statistical processing and the excess or deficiency code amount generated in the encoding of the previous block. The allocated code amount for the block is finally determined based on the accumulation of , and is output to the coding abort unit 11 . Each time a Huffman code is generated, the coding abort unit 11 monitors whether the allocated code amount is exceeded when the code and EOB are sent. If it does not exceed
The generated code amount is subtracted from the allocated code amount.

[0044] If the amount exceeds the number, the variable length encoding unit 4 is controlled to stop the encoding of that block, add EOB without outputting the code, and use the generated code amount (EOB only) as the allocated code amount. Subtract from. However, for the DC component, EOB is added after its code is also output.

At the position where the above-mentioned special code is inserted, the bit stuff section 5 monitors whether the code amount is in byte units, and if it is not aligned in byte units, the bit "1" is set.
” are added as many times as necessary.

[0046] The byte stuffing unit 6 divides the code into bytes including the bit stuff and writes it in hexadecimal notation as "
When the sequence "FF" is found, "00" is inserted immediately after, and the increase in the code amount is transmitted to the coding abort section 11 and the code amount calculation section 8. The increase in code amount is subtracted from the allocated code amount before monitoring the code amount of the code following the code that caused bit stuffing or byte stuffing.

As described above, since Huffman encoding is performed while monitoring the allocated code amount, when the coding of one block is completed (including the termination of coding), the allocated code amount is "0" or The value becomes negative when the allocated code amount is exceeded due to encoding of only the DC component or occurrence of bit stuffing or byte stuffing.

[0048] When the encoding of one block is finished,
If the allocated code amount is surplus, it will be added to the surplus or deficiency code amount,
It is used to determine the allocated code amount for the next block. Furthermore, if the allocated code amount is insufficient, the code amount is allocated from the code amount secured in advance as a margin from the target code amount. If this margin is used up, it is subtracted from the surplus/deficiency code amount and used to determine the allocated code amount for the next block.

[0049] In addition to the above, when encoding the last block of the entire image, a margin is secured, and the allocated code amount can be reduced by encoding the DC component or by bit stuffing or byte stuffing. If it exceeds, this margin will be used.

The Huffman code table used in Huffman encoding is generated by the variable length code table generation unit 3 by summing up the frequency of occurrence of the transform coefficient values of the quantized block image data output from the quantization unit 2. be done. Alternatively, a specific Huffman code table may be used regardless of the frequency of occurrence, but by adding the EOB code as described above, it is assumed that byte stuffing will not occur. This ensures that the generated code amount falls within the allocated code amount by terminating the encoding.

The image data encoded in this manner is multiplexed with the finally used quantization table, Huffman code table, and other additional information in the code output section 7, and is outputted from the output terminal B.

In this embodiment, the block size is 8×8, and the orthogonal transformation is DC
Although Huffman codes were used for T (discrete cosine transform) and variable length encoding, the embodiments of the present invention are not limited to these.

According to the image data compression encoding device and encoding method of the present invention, image data is divided into blocks, orthogonal transformation is performed for each divided block, and the output of this transformation is quantized. When encoding image data by variable-length encoding the encoded output, a first encoding process is performed using a provisional quantization width, and the quantization width is optimized and Predict the allocated code amount for each block and use the optimized quantization width,
In addition, a specific variable-length code is assigned to the EOB, and two types of margin code amounts are allocated in advance from the target code amount for the final block and intermediate blocks. By performing the second encoding while aborting the encoding, in order to guarantee compatibility with the proposed international standard system, we can deal with byte stuffing while maintaining a simple configuration within a certain processing time.
Image data can be encoded within a target code amount and with almost no deterioration in image quality.

As described in detail above, in this embodiment, the image data can be encoded within a certain period of time because the processing is completed by two-stage encoding, that is, statistical processing and encoding processing. In addition, by optimizing the quantization width and allocating the code amount for each block through statistical processing, it is possible to not only control the code amount accurately relative to the target code amount, but also to prevent the occurrence of accidental byte stuffing. However, by assigning a specific code to the EOB code and providing two types of margins, it is possible to suppress the amount of generated code within the target code amount while suppressing deterioration of image quality. Further, the present invention is not limited to the embodiments described above, and it goes without saying that various modifications and applications can be made without departing from the gist of the invention.

[0055]

Effects of the Invention As described in detail above, according to the present invention, it is possible to maintain image quality while ensuring compatibility with the international standard system, and to achieve the desired amount of code within a certain processing time with a simple configuration. It is possible to provide an encoding device and an encoding method for image data that encodes image data so that it fits within

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the configuration of an image data compression encoding device as an embodiment of the present invention.

FIG. 2 is a block diagram for explaining the encoding method of the image data compression encoding apparatus having the configuration shown in FIG. 1;

FIG. 3 is a diagram showing a scanning order for searching for transform coefficients of quantized block image data.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1... Orthogonal transformation unit, 2... Quantization unit, 3... Variable length code table generation unit, 4... Variable length encoding unit, 5... Bit stuff unit, 6... Byte stuff unit, 7... Code output unit, 8... Code amount calculation unit, 9... Quantization width prediction unit, 10... Code amount allocation unit,
11... Encoding abort control unit 11, A... Input end, B... Output end.

Claims (2)

[Claims] 1. An image data code in which image data is divided into a plurality of blocks, orthogonal transformation is performed for each block, the transformed output is quantized by a quantization means, and variable length encoded by a variable length encoding means. In the encoding device, a code amount calculation means calculates the code amount for each block of the variable-length encoded transform coefficients and also calculates the code amount of the entire image, and the entire image calculated by the code amount calculation means. a quantization width prediction means for predicting an optimal quantization width from the code amount and a target code amount; a code amount for each block calculated by the code amount calculation means; and prediction by the quantization width prediction means. code amount allocation means that allocates a code amount to each block from the quantization width that has been determined; and a variable length code table that allocates a specific code to an end of block (EOB) code indicating the end of the block. variable-length code table generation means for generating a variable-length code table; excess/deficiency code amount calculation means for calculating the accumulation of excess/deficiency of code amount caused by encoding of each block before the block to which the EOB code is added; An allocated code amount for determining the allocated code amount of the block from the surplus/deficit code amount calculated by the code amount calculation means, the allocated code amount, and the margin code amount allocated from the preset target code amount. a determining means, a coding aborting means for aborting the variable length encoding so as not to exceed the allocated code amount determined by the allocated code amount determining means, and a code output from the variable length encoding means and other additions. A first encoding process is performed in which the code output means for multiplexing and outputting information and the quantization means perform quantization using a preset temporary quantization width, and as a result, 1. An image data encoding device comprising: a control means for performing control so as to perform a second encoding process using the obtained optimal quantization width and allocated code amount. 2. An image data code in which image data is divided into a plurality of blocks, orthogonal transformation is performed for each block, the transformed output is quantized by a quantization means, and variable length encoded by a variable length encoding means. In the quantization method, the first step is to encode the image data and check the amount of code for each block and the amount of code for the entire image, and determine the optimal quantization width and each amount from the information obtained in the first step. A second step of predicting the allocated code amount for each block, a third step of performing quantization again using the quantization width predicted in the second step, and an allocated code amount predicted in the second step. and a fourth step of determining the allocated code amount for the block based on the excess or deficiency of the code amount that occurred before that block and the code amount of the margin allocated in advance from the target code amount. A method for encoding image data, comprising a fifth step of performing variable length encoding so as to fit within the determined allocated code amount.