JPH0537783A - Coder for picture data - Google Patents

Abstract

PURPOSE:To attain circuit integration without making the circuit configuration of the data compression system coder for code quantity control complicated by adopting other integrated circuit chip for a quantization width decision means than that for a means required for other coding processing. CONSTITUTION:A picture data subjected to block processing from a memory 50 is converted in an orthogonal transmission circuit 61. The orthogonal transformation coefficient output is given to a quantization circuit 62, in which the signal is subjected to linear quantization with a quantization width for each frequency component set in advance. The quantized transformation coefficient is given to an entropy coding circuit 63, in which the data is coded and the resulting picture data is given to a code output circuit 64, in which the data is segmented while a bus width is arranged for output control to a recording medium 70. A quantization coefficient prediction circuit 65 predicts an optimum quantization coefficient based on a setting object code quantity and a total code quantity from a system controller 80 and outputs the result to the circuit 62. Then only the circuit 65 is separated from the other circuits and integrated into a separate IC 60B.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding apparatus for highly compressing and coding image data into a predetermined capacity.

[0002]

2. Description of the Related Art Generally, when an image signal picked up by a solid-state image pickup device typified by a CCD (charge coupled device) is recorded as digital data on a recording medium such as a memory card, a magnetic disk, or a magnetic tape, The amount of data will be enormous. Therefore, in order to record many frame images in a limited recording capacity range, it is usually necessary to perform some high-efficiency compression on the obtained image signal data.

Further, in a digital electronic still camera or the like, the captured image is stored as digital data in a recording medium such as a memory card, a magnetic disk, or a magnetic tape instead of the silver salt film, so that one image is stored. The number of images that can be recorded on a memory card, a magnetic disk, or one roll of magnetic tape is limited, and it is necessary to guarantee recording of the limited number of images. Moreover, in this case, it is also necessary that the time required for recording / reproducing data is short and constant.

Similarly, when recording a moving image in a digital VTR (video tape recorder), a digital moving image file, etc., a predetermined amount of frames can be recorded without being affected by the image data amount per frame. There must be.

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 the data recording / reproducing process must be short and constant. There is.

As a highly efficient compression method for image data, an encoding method combining orthogonal transform encoding and variable length encoding is widely known. As a typical example, there is a method being studied in the international standardization of still image coding.

An outline of this method will be described below. First, the image data is divided into blocks of a predetermined size, and each divided block is a two-dimensional D as an orthogonal transformation.
Perform CT (discrete cosine transform). Next, linear quantization is performed according to each frequency component, and Huffman coding is performed on the quantized value as variable length coding. At this time, as for the DC component, the difference value from the DC component of the neighboring block is Huffman-encoded. The AC component scans from a low frequency component called a zigzag scan to a high frequency component, and a two-dimensional Huffman scan is performed based on the number of consecutive invalid (value "0") components and the value of the effective component that follows. Encode.

Since Huffman coding, which is variable-length coding, is used only in this basic portion, the code amount is not constant for each image. Therefore, the following method has been proposed as a method of controlling the code amount.

First, at the same time as the processing of the basic part is performed, the total code amount generated in the entire screen is obtained. From this total code amount and the target code amount, the optimum quantization width for approaching the target code amount for the DCT coefficient is predicted. Next, the process after the quantization of the basic part is repeated using this quantization width. Then, based on the total code amount generated this time, the total code amount generated last time, and the target code amount, the optimum quantization width for approaching the target code amount again is predicted. Then, when the predicted quantization width and the previous quantization width are sufficiently close to each other and the total code amount generated this time is smaller than the target code amount, the processing is ended and the code is output. If not, the process is repeated using the new quantization width.

For the quantization width, a quantization matrix, which is a basic form showing relative quantization characteristics for each frequency component, is prepared, and this quantization matrix is multiplied by a quantization coefficient α to obtain a required quantization width. To get Specifically, first, the above-mentioned basic part is quantized by the quantization width obtained by using the standard quantization coefficient α, and this is variable-length coded, and the total code amount obtained by this is The information is compared with a preset target total code amount that is a limit to be stored, and when the target total code amount is reached, a final encoding process is performed using the quantization width. If the total amount of code does not fall within the target total code amount, the optimum quantization coefficient α for approaching the target total code amount is obtained by linear prediction, for example, from the generated total code amount and the target total code amount. A more optimized quantization width is calculated from the quantization coefficient α and the quantization matrix, and the final encoding processing is performed using this. The quantization width is changed by such a method.

The above operation will be specifically described with reference to FIG. First, as shown in (a), one frame of image data (the image of one frame exemplified in the international standardization proposal is 720 × 576 pixels) is divided into blocks of a predetermined size (for example, 8 × 8 pixels). Blocks A, B, C
...), and as shown in (b), two-dimensional DCT is performed as an orthogonal transform for each of the divided blocks.
Sequentially store on a × 8 matrix. The image data is
When viewed in a two-dimensional plane, it has a spatial frequency that is frequency information based on the distribution of grayscale information.

Therefore, by performing the above DCT,
Image data is converted into DC component AC and AC component AC,
Data indicating the value of the DC component DC is located at the origin position ((0,0) position) on the 8 × 8 matrix, and the maximum frequency value of the AC component AC in the horizontal axis direction is located at the (0,7) position. And data indicating the maximum frequency value of the AC component AC in the vertical axis direction at the (7,0) position,
Data indicating the maximum frequency value of the AC component AC in the diagonal direction is stored in the (7, 7) position, respectively. At the intermediate position, the frequency data in the direction associated with each coordinate position is stored in such a manner that the frequency data having a frequency higher than the origin side appears.

Next, the stored data at each coordinate position in this matrix is divided by the quantization width for each frequency component to perform linear quantization according to each frequency component ((c in FIG. 3). )), Huffman coding is performed as variable length coding on the quantized value. At this time, regarding the DC component DC, the difference value from the DC component of the neighboring block is expressed by a group number (number of additional bits) and additional bits, and the group number is Huffman coded, and the obtained codeword and additional bits are Together, they are coded data (Fig. 3
(D1), (d2), (e1), (e2)).

Coefficients (values that are not "0") valid for the AC component AC are represented by a group number and additional bits. Therefore, the AC component AC performs a scan from a low frequency component to a high frequency component called a zigzag scan, and a continuous number of invalid (value “0”) components (run number of zero) is followed by a valid number. Two-dimensional Huffman encoding is performed from the group number of the component value, and the obtained codeword and additional bits are combined to form encoded data.

In Huffman coding, the peak of the occurrence frequency in each data distribution of the DC component DC and the AC component AC per frame image is the center, and the number of data bits is reduced toward the center, and the periphery is reduced. The data is encoded in the form of bit allocation to obtain a codeword by increasing the number. The above is the basic part of this method.

Since only the basic part uses the Huffman coding which is the variable length coding, the code is not constant for each image. Therefore, as a method of controlling the code amount, for example, the following processing is performed. To do.

First, when the temporary quantization coefficient α is used, the processing of the basic portion is performed with the quantization width for each frequency component obtained by multiplying the quantization coefficient α and the quantization coefficient α determined. At the same time, the total code amount (total number of bits) generated on the entire screen is obtained ((g) in FIG. 3). Next, from this total code amount, the target code amount, and the provisional quantization coefficient α used, etc., the optimum quantization coefficient α for approaching the target code amount for the DCT coefficient is predicted (Fig. (H) in 3), using the quantized coefficient α ((i) in FIG. 3), the processes after the quantization of the basic part described above are repeated. Then, from the total code amount generated this time, the total code amount generated last time, the target code amount, the quantization coefficient α used this time, and the quantization coefficient α used the previous time, the target code amount is again set. The optimum quantization coefficient α for approaching is Newton RaphsonIterration (Newton RaphsonIterration
).

When the predicted quantized coefficient α and the previously quantized coefficient α are sufficiently close to each other and the total code amount generated this time is smaller than the target code amount, the processing is terminated. The encoded data generated this time is output and stored in the memory card ((f) in FIG. 3). If not, change the quantized coefficient α and change this new quantized coefficient α
Repeat the process with.

[0019]

When the above-described encoding is used for compressing a video signal in a device such as an electronic still camera or a video camera, the circuit for performing the encoding is realized as an integrated circuit (IC). It is necessary to.

However, it is difficult to realize a part of the above-mentioned compression processing as an IC because the circuit becomes extremely complicated.

That is, in the Newton-Raphson iteration, which is a prediction method of the optimum quantized coefficient α, the quantized coefficient α is given by the following equation (1), for example. α = 10 ^{{1.7−1.5 (log} 10 ^BR−log 10 ^br)} (1) However, in the equation (1), BR is a target code amount (per pixel), br Is the actually obtained code amount (per pixel).

As can be seen from the above equation (1), the prediction of the quantization coefficient α requires a complicated operation such as logarithmic exponentiation. If such an operation circuit is to be integrated in the IC, the chip area will be increased. Was increased, or the processing time became very long, which made it difficult to realize.

The present invention has been made in view of the above points, and realizes an encoding device for executing a data compression method, which is being considered as an international standard, as an integrated circuit without complicating the circuit configuration. The purpose is.

[0024]

In order to achieve the above object, an image data encoding apparatus according to the present invention comprises:
Receiving means for receiving information for determining a quantization width from the outside, quantizing means for quantizing an orthogonal transformation output of image data with a quantization width corresponding to the information received by the receiving means, and the quantizing means. Variable length coding means for variable length coding the quantized output of the means, total code amount calculating means for calculating the total code amount generated by the variable length coding means, and total code amount calculating means It is characterized in that the output means for outputting the total code amount to the outside is integrated on the same integrated circuit chip.

Further, the integrated circuit chip is further provided with an orthogonal transformation means for orthogonally transforming image data for each block of a predetermined size.

Further, an optimum quantization width is obtained using the set target code amount and the total code amount output from the output means, and information corresponding to the obtained optimum quantization width is sent to the receiving means. It is characterized in that the quantizing width determining means to be output to this is integrated in an integrated circuit chip different from the integrated circuit chip.

[0027]

In the image data encoding apparatus according to the present invention, the image data is orthogonally transformed for each block, the transformed output is quantized using a certain quantization width, and then the quantized output is variable length. By encoding and setting the code amount and the process of changing the quantization width using the total code amount generated in the encoding, control is performed so that the total code amount falls within the set target code amount. Perform data compression. In this case, at least a receiving means for receiving information for determining a quantization width, for example, quantization coefficient information, a quantizing means for quantizing an orthogonal transform output, and a variable length coding for variable length coding this quantized output. The means, the total code amount calculating means for calculating the total code amount generated in the encoding, and the output means for outputting the calculated code amount are configured to be integrated on the same integrated circuit chip, and are calculated. The total code amount is output to a quantization width determining unit that obtains an optimum quantization width using the set target code amount and the generated total code amount, and the optimization amount obtained by this quantization width determining unit is also output. Encoding is performed by receiving the quantized coefficient information corresponding to the generated quantum width. That is, since the quantization width determining means is integrated in a different integrated circuit chip from the means necessary for other encoding processing to perform complicated quantization coefficient prediction calculation, it is considered as an international standard. It is possible to realize an encoding device that executes the existing data compression method as an integrated circuit without complicating the circuit configuration.

[0028]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a coding circuit as an embodiment of a coding device of the present invention, and FIG. 2 is a block diagram of a digital electronic still camera to which this coding circuit is applied. In the following description, a mechanism of a digital electronic still camera which is not directly related to the present invention will be omitted from illustration and description.

Digital electronic still camera body 100
2, the lens 10, the image pickup device 20 such as a CCD (charge coupled device), the process circuit 30, the A / D
An (analog / digital) converter 40, a memory 50, an encoding circuit 60, a recording medium 70, and a system controller 80 are provided.

The lens 10 is placed in front of the image sensor 20 and projects a subject image onto the image sensor 20. Image sensor 2
The video signal output by 0 is the process processing circuit 30.
After being subjected to color separation, white balance, gamma correction, and the like, the A / D converter 40 converts the signals into digital signals. The digitally converted image data is output to the memory 50. The memory 50 is, for example, a RAM
(Random access memory). Memory 50
When the data is read from the encoder to the encoding circuit 60, a block forming process for dividing into 8 × 8 blocks is performed by memory address control. The encoding circuit 60 performs compression encoding processing on the input image data, and the processed data is recorded on the recording medium 70. The system controller 80 is composed of, for example, a microcomputer and controls each of the above-mentioned components.

As shown in FIG. 1, the coding circuit 60 includes an orthogonal transformation circuit 61, a quantization circuit 62, an entropy coding circuit 63, a code output circuit 64, and a quantization coefficient prediction circuit 6.
5, a code amount calculation circuit 66, a quantization coefficient input circuit 67, a code amount output circuit 68, and a control circuit 69.

The orthogonal transformation circuit 61 performs orthogonal transformation on the block-processed image data output from the memory 50 for each block, here, DCT (discrete cosine transform) as an example. Quantization coefficient input circuit 67
Is an 8-bit input port and receives a quantized coefficient value from the quantized coefficient predicting circuit 65. The quantization circuit 62 performs linear quantization on the orthogonal transform coefficient output from the orthogonal transform circuit 61 using a quantization width preset for each frequency component for each frequency component. The entropy coding circuit 63 as a variable length coding circuit entropy codes the transform coefficient quantized by the quantization circuit 62. In this embodiment, Huffman coding is used. The code output circuit 64 controls the image data coded by the entropy coding circuit 63 so that the image data is divided into uniform bus widths, for example, 8 bits, and the divided data is output to the recording medium 70.

The code amount calculation circuit 66 calculates the code amount of the image data coded by the entropy coding circuit 63 and outputs the calculation result to the code amount output circuit 67.
The code amount output circuit 67 divides the total code amount into 8 bits,
Output to the quantization coefficient prediction circuit 65. The control circuit 69 is
Each component of the encoding circuit 60 is controlled.

The quantized coefficient predicting circuit 65 predicts the optimum quantized coefficient using the target code amount set by the system controller 80 and the obtained total code amount, and outputs it to the quantized coefficient input circuit 67. It outputs to the quantization circuit 62 via. The quantized coefficient prediction circuit 65 can be configured by, for example, a floating point processor.

In the encoding circuit 60 having the above configuration, the orthogonal transformation circuit 61, the quantization circuit 62, the entropy encoding circuit 63, the code output circuit 64, the code amount calculation circuit 66, the quantization coefficient input circuit. 67, a code amount output circuit 68,
The control circuit 69 and the control circuit 69 are integrated in one IC 60A, and the quantization coefficient prediction circuit 65 is integrated in another IC 60B different from the IC 60A.

Next, the operation of the digital electronic still camera configured as described above will be described.

Prior to photographing an image, the system controller 80 provides the quantization coefficient prediction circuit 65 with target code amount information. Here, the target code amount may be fixed to a value unique to the camera, or a desired value may be selected each time by a changeover switch.

Next, when photographing is started by pressing a shutter (not shown), the subject image projected by the lens 10 is converted into a video signal by the image pickup device 20 and output to the process circuit 30. In this process circuit 30, color separation, white balance,
Processing such as gamma correction is performed. The video signal processed as described above is converted into a digital signal by the A / D converter 40 and output to the memory 50 as a luminance and color difference line sequential signal. The memory 50 stores 8 × video signals.
While performing the blocking process to divide into 8 blocks,
The luminance signal and then the color difference signal are output to the encoding circuit 60 in this order.

In the encoding circuit 60, the orthogonal transformation circuit 61 first performs DCT on the input image data. Next, the obtained DCT transform coefficient is linearly quantized by the quantization circuit 62 by using the quantization width for each frequency component for each frequency component. The quantization width here corresponds to the target code amount information previously given to the quantization coefficient prediction circuit 65 from the system controller 80,
The quantization width corresponding to the quantization coefficient α set by the quantization coefficient prediction circuit 65 and given to the quantization circuit 62 via the quantization coefficient input circuit 67 is used.

The transform coefficient quantized by the quantization circuit 62 in this way is Huffman coded by the entropy coding circuit 63. At this time, regarding the DC component DC, the difference value from the DC component of the neighboring block is expressed by a group number (number of additional bits) and additional bits, and the group number is Huffman coded, and the obtained codeword and additional bits are Together, they are coded data. For the AC component AC, zigzag scanning (scanning from low frequency components to high frequency components) is performed, and the number of consecutive invalid components (number of runs of zero) and the group number of the value of the effective components that follow are identified. Is subjected to two-dimensional Huffman coding, and the obtained codeword and additional bits are combined to form coded data. The code amount calculation circuit 66 calculates and stores the code amount of the encoded image data.

The above processing is sequentially performed for each block. Then, when the processing of the blocks for the entire screen is completed, the generated total code amount is output from the code amount calculation circuit 66 to the quantization coefficient prediction circuit 65 integrated in another IC 60B via the code amount output circuit 68. To be done.

In the quantization coefficient prediction circuit 65, the above (1)
The optimum quantized coefficient α is predicted based on the equation. Along with this, the total code amount is compared with the target code amount, and if the total code amount is less than or equal to the target code amount and the difference between the previous quantized coefficient and the new quantized coefficient is within the set value, this Is output to the system controller 80.

In response to this, the system controller 80 starts the operation of writing the encoded data in the recording medium 70.

That is, again from the memory 50 to the encoding circuit 60.
The image data subjected to the block processing is input to and the DCT is performed for each block by the orthogonal transform circuit 61. The quantization circuit 62 linearly quantizes the obtained DCT transform coefficient for each frequency component. However, this time, the quantization width corresponding to the previously predicted optimal quantization coefficient α is used. The quantized transform coefficient is Huffman coded by the entropy coding circuit 63.
In this case, as in the previous case, the difference value between the DC component DC and the DC component of the neighboring block is encoded, and the AC component AC is two-dimensionally Huffman-encoded. The encoded image data is output from the entropy encoding circuit 63 to the code output circuit 64, where it is divided into eight bits and output to the recording medium 70.

On the other hand, in the quantization coefficient prediction circuit 65, the optimum quantization coefficient α is calculated from the total code amount and the target code amount as described above.
As a result of predicting using the equation (1), when the total code amount is not within the target code amount, or when the difference between the previous quantized coefficient and the new quantized coefficient is larger than a predetermined value, the optimum quantum Processing for predicting the conversion coefficient α is performed.

That is, again, from the memory 50 to the encoding circuit 6
Blocked image data is input to 0, and the orthogonal transform circuit 61 performs DCT for each block. The quantization circuit 62 is applied to the obtained DCT transform coefficient.
By this, linear quantization is performed for each frequency component, but here, the quantization width corresponding to the optimum quantization coefficient α obtained in the previous encoding is used.

The quantized transform coefficient is Huffman coded by the entropy coding circuit 63. Similarly to the above, for the DC component DC, the difference value from the DC component of the neighboring block is encoded, and for the AC component AC, two-dimensional Huffman encoding is performed. Then, the code amount for the AC component of each block is integrated by the code amount calculating circuit 66, and when the processing of the blocks for the entire screen is completed, the total code amount generated is calculated from the code amount calculating circuit 66 by the quantization coefficient. It is output to the prediction circuit 65. In the quantization coefficient prediction circuit 65, the total code amount is compared with the target code amount, and if the total code amount is within the target code amount, a signal indicating this is output to the system controller 80, which is described above. The writing operation to the recording medium 70 is performed as described above. If the generated total code amount is not within the target code amount, the process for predicting the optimum quantized coefficient α is repeated.

As described above, according to the present invention, the quantized coefficient prediction circuit is integrated in the integrated circuit chip different from the circuit required for other coding processing, thereby performing the complicated quantized coefficient prediction calculation. Can be realized. As a result, it is possible to integrate an encoding device that encodes any image data so as to be within a certain code amount in an IC without complicating the circuit.

In the above embodiment, all the circuits constituting the encoding circuit 60 except the quantization coefficient predicting circuit 65 are integrated in one IC, but the present invention is not limited to this. It is not something that will be done. For example, the orthogonal transformation circuit 61 is integrated into one IC, the quantization circuit 62, the entropy coding circuit 63, the code output circuit 64, the quantization coefficient input circuit 67, the code amount calculation circuit 66, the code amount output circuit 68, and the control circuit. 69 may be integrated in another IC.
The system controller 80 may also serve as the quantization coefficient prediction circuit 65.

In the above embodiment, the quantization coefficient is IC
However, the quantization width itself corresponding to each frequency component may be given. Alternatively, the quantization width may be set in advance in the quantization circuit 62 so as to be switched in several steps, and this switching selection information may be given.

[0052]

As described above in detail, according to the present invention, the circuit configuration of the coding apparatus for executing the data compression method which is considered as the international standard and which controls the code amount by optimizing the quantization width is provided. Can be realized as an integrated circuit without complicating.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of an encoding circuit as an embodiment of an image data encoding device according to the present invention.

FIG. 2 is a block configuration diagram of a digital electronic still camera to which the encoding circuit of FIG. 1 is applied.

FIG. 3 is an operation transition diagram for explaining the principle of a conventional compression method.

[Explanation of symbols]

50 ... Memory, 60 ... Encoding circuit, 61 ... Orthogonal transformation circuit, 62 ... Quantization circuit, 63 ... Entropy coding circuit, 64 ... Code output circuit, 65 ... Quantization coefficient prediction circuit,
66 ... Code amount calculation circuit, 67 ... Quantization coefficient input circuit, 6
8 ... Code amount output circuit, 69 ... Control circuit, 70 ... Recording medium, 80 ... System controller.

Claims (3)

[Claims] 1. A receiving means for receiving information for determining a quantization width from the outside, and a quantizing means for quantizing an orthogonal transformation output of image data with a quantization width corresponding to the information received by the receiving means. Variable length coding means for variable length coding the quantized output of the quantizing means, total code amount calculating means for calculating the total code amount generated by the variable length coding means, and total code amount calculation An encoding device for image data, characterized in that an output means for outputting the total code amount calculated by the means to the outside is integrated in the same integrated circuit chip. 2. The image data coding apparatus according to claim 1, wherein the integrated circuit chip is further provided with orthogonal transformation means for orthogonally transforming image data for each block of a predetermined size. .. 3. An optimum quantization width is obtained using the set target code amount and the total code amount output from the output means,
The quantizing width determining means for outputting information corresponding to the obtained optimum quantizing width to the receiving means is integrated in an integrated circuit chip different from the integrated circuit chip. The described image data encoding device.