JP4041245B2 - Image encoding device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image encoding apparatus that compresses and encodes image data, and more particularly to an image encoding apparatus that keeps the code amount within the limit while keeping the image quality as good as possible when the output code amount is limited.
[0002]
[Prior art]
Conventionally, JPEG is known as a standard of an image compression encoding method using discrete cosine transform. In this method, image data is subjected to discrete cosine transform for each block, and the transform value is quantized. Discrete cosine transform classified as transform coding is a well-known transform method for transforming image data into the spatial frequency domain, and is widely used for image data because of its high ability to reduce information entropy.
[0003]
Discrete cosine transform and quantization causes the image to undergo irreversible distortion and reduce information entropy. Since the discrete cosine transform and JPEG quantization are integrated in operation, they can be collectively regarded as a transform coding unit. After this, entropy coding is further performed. Entropy encoding is performed so that the output code approaches information entropy, and distortion due to encoding is not introduced. The part that implements this is a coefficient encoding unit.
[0004]
In the coefficient encoding unit, among the coefficients obtained by quantization, the DC coefficient is obtained as a difference from the DC coefficient value of the previous block, and the difference value is Huffman encoded. On the other hand, AC coefficients are scanned in a zigzag order and are two-dimensionally Huffman encoded as a combination of the number of zeros and non-zero coefficients. Since both the discrete cosine transform and the quantization increase the appearance frequency of zero in the AC coefficient, a Huffman code is assigned to a combination of a continuous number of zeros and a non-zero coefficient. Therefore, the coefficient encoding unit finds a non-zero coefficient or outputs a code when the end of the block is reached. Although an image compressed according to JPEG is accompanied by deterioration, by appropriately setting a quantization table used for quantization, significant data compression is possible while maintaining a good subjective evaluation value.
[0005]
Conventionally, a video conference codec for video conferencing is known for the purpose of image communication for transmitting images that change from moment to moment. A moving image codec has a high compression rate and is excellent, but requires a dedicated LSI and must be operated with a high-speed clock, resulting in an increase in power consumption. In a device that needs to reduce power consumption as much as possible, such as a portable communication device, image transmission can be performed using a JPEG codec instead of a moving image codec. The JPEG processing method is simpler than a moving image codec, and can be realized by a CPU and software without using a dedicated LSI. One of the reasons why the amount of calculation of the moving image codec is large is that the difference between frames is encoded and transmitted. At this time, the encoder uses a local decoder to display the same image as that decoded by the other party. It must be generated. On the other hand, if the JPEG codec is not used and the inter-frame difference is not used, the required calculation performance is lowered, and there is an advantage that power consumption can be suppressed.
[0006]
However, since the JPEG encoder cannot achieve a compression rate as high as that of a moving image codec, it is necessary to reduce the amount of generated code at the expense of image size, image quality, frame rate, and the like.
[0007]
As a conventional image coding apparatus for controlling the code amount, a transmission rate control method in image coding of Japanese Patent No. 2647272 is known, which is shown in FIG. According to this, the average code amount and the instantaneous code amount per unit time are calculated by the rate calculation circuit 205 and notified to the VLC control circuit 204. When these rates exceed the set values, the VLC control circuit 204 The code of the coefficients up to the middle is sent, and the sign of the remaining coefficients is not forcibly sent. Thereby, control of the code amount is realized. At this time, since the characteristics of the quantizer 202 are not changed, there is no need to consider whether or not the code amount is controlled on the decoding side, and there is an advantage that a general-purpose and simple decoder can be used. In addition, since the coefficient zero replacement is performed from a coefficient on the high frequency side, there is an advantage that the image quality is not greatly impaired even when the code amount is limited.
[0008]
Conventionally, an image coding apparatus that performs coding by controlling a quantizer in accordance with the position of a block in an image to reduce important portions with high accuracy and reduce information on unimportant portions has been disclosed in Japanese Patent Application Laid-Open No. Hei 4- It is known as a 247789 encoder and decoder.
[0009]
[Problems to be solved by the invention]
However, in the conventional image coding apparatus that controls the code amount of Japanese Patent Publication No. 2647272, there is a problem that the code amount control with a difference in image quality cannot be performed in consideration of the position in the image. The important part of the image is usually near the center and less important as the periphery. When it is necessary to reduce the amount of code at the expense of image quality, the image quality deterioration in the peripheral portion can be tolerated, but the image quality deterioration in the central portion should be avoided. For this reason, it is desirable to allocate a large amount of code to the block located at the center of the image and assign a small amount of code to the block located around the image, but this has not been possible with the conventional image coding apparatus. The reason for this is that in the conventional apparatus, the criterion for truncating the block coefficient does not consider the position of the block in the image.
[0010]
On the other hand, the conventional image coding apparatus disclosed in Japanese Patent Laid-Open No. 4-247789 which controls the quantizer according to the position of the block overcomes this problem, but the decoder has a dedicated decoder corresponding to the encoder. The JPEG decoder that is widely used cannot be used and the image can be transmitted. However, when the transmitted image is used for other purposes or attached to an e-mail. Has the problem that it must be recompressed with a JPEG encoder. Further, although this image coding apparatus can change the weighting according to the block position, it does not have a code amount control function for keeping the generated code amount within a certain limit. That is, although the balance of the code amount of each block can be controlled, there is a problem that the total code amount obtained by adding them can not be controlled.
[0011]
[Means for Solving the Problems]
In order to solve the above problem, an image coding apparatus according to claim 1 is a transform coding means for discrete cosine transforming and quantizing image data for each block and outputting coefficients, and the transform coding means outputs In an image encoding device comprising coefficient encoding means for encoding a coefficient,
An allowable number storage means for storing the allowable number of code outputs of the coefficient encoding means;
Spatial allowable number increase means for increasing the allowable number of times stored in the allowable number storage means according to the position of the image data in the image;
Spatial threshold value specifying means for specifying a threshold value of the coefficient according to the position of the image data in the image,
The coefficient encoding means compares the spatial frequency of the coefficient output from the transform encoding means with the threshold specified by the spatial threshold specifying means, and encodes the coefficient when the spatial frequency of the coefficient is lower than the threshold. If the allowable number stored in the allowable number storage means is positive, the coefficient is encoded and the code is output.If the allowable number is not positive, the coefficient is output as zero. Without
Further, when the coefficient encoding means outputs a code, the allowable number of times stored in the allowable number storage means is controlled to be decreased by the number of the outputted codes.
[0012]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a preferred embodiment of the present invention. FIG. 2 is a flowchart showing the procedure of the operation.
[0013]
The image coding apparatus of the present invention operates based on the position of an MCU (Minimum Coded Unit) image. The MCU is an encoding unit including a plurality of blocks, and includes a luminance component block and a color difference component block, and is the minimum encoding unit as a color image. Since the order in which the image coding apparatus processes the MCU is fixed, the serial number is determined for the MCU, and the MCU position can be specified by counting this with the MCU counter. Each processing unit is notified of the value of the MCU counter, so that processing according to the position of the MCU in the image can be performed.
[0014]
When the processing is started in FIG. 2, first, an MCU counter for designating an MCU is initialized to 0 and an allowable number storage means 105 is initialized to 0 in processing S01.
[0015]
A subsequent process S02 is a determination process for checking whether all the MCUs constituting the image have been processed. When the determination result is Yes, the processing for one image is completed. When the determination result is No, the process proceeds to step S03.
[0016]
In step S03, the value of the MCU counter is notified to the spatial allowable number increase means 104. Spatial allowable number increase means 104 stores the increment value of the allowable number corresponding to the MCU number. When receiving the notification in step S03, the spatial allowable number increase means 104 increases and updates the stored value of the allowable number storage means 105 using this stored value. Further, the value of the MCU counter is also notified to the spatial threshold value specifying means 103. The spatial threshold designating unit 103 sets a threshold according to the position of the MCU in the image.
[0017]
Subsequently, in process S04, the image data of the MCU specified by the MCU counter is input to the transform coding unit 101 in units of 8 × 8 pixel blocks. A single MCU has a plurality of blocks, and the blocks are processed in order. In step S04, the transform coding unit 101 performs two-dimensional discrete cosine transform on the input block, and further performs quantization. Quantization conforming to JPEG is a process of multiplying a coefficient value obtained by discrete cosine transform by an inverse number of a quantization step and rounding to an integer. Since the quantization operation is multiplication, it is integrated with discrete cosine transform that executes a large amount of multiplication, and both are used as transform coding section 101.
[0018]
In the quantization table, a quantization step size is set in advance according to the properties of the input image and the image quality and code amount of the image output from the image encoding apparatus. As is well known, an 8 × 8 coefficient block output by discrete cosine transform is a coefficient in the spatial frequency domain, and the position of each coefficient in the block and the spatial frequency have a one-to-one correspondence. The quantization step size is small for the low frequency side coefficient and large for the high frequency side coefficient, so that a large amount of information is allocated to the low frequency side coefficient and a small amount of information is allocated to the high frequency side coefficient. be able to. The quantization table can be switched depending on whether the data block is a luminance component or a color difference component of the image, and a quantization step suitable for each color component having a different statistical property can be designated. .
[0019]
In the process S05, the coefficient output from the transform encoding unit 101 is input to the coefficient encoding unit 102. The coefficient encoding unit 102 receives the quantized 8 × 8 coefficient block and performs entropy encoding on this.
[0020]
When processing of a plurality of blocks included in the MCU is completed, the process proceeds to process S06, the MCU counter is incremented, and the process returns to the determination process S02.
[0021]
Next, it is the coefficient encoding unit 102 that performs code amount control in consideration of the spatial position, which is a feature of the present invention. FIG. 3 is a flowchart showing the operation, which will be described based on this.
[0022]
In response to the input of the coefficient block, the coefficient encoding unit 102 encodes the DC coefficient and outputs a code in step S11. For the DC coefficient, the difference from the previous DC coefficient of the same color component is obtained, the difference value is decomposed into the level value to which it belongs and the remaining bits within that level, the level value is Huffman encoded, and then the remaining bits are Attach.
[0023]
When the coefficient encoding unit 102 encodes the AC coefficient, the blocks are scanned zigzag in order from the low frequency to the high frequency, and the coefficient that is not zero is searched for while counting the number of coefficients having the value zero.
[0024]
Here, the order of zigzag scanning is shown in FIG. FIG. 4 shows an 8 × 8 coefficient block that is a result of the transform coding unit 101 performing discrete cosine transform and quantization, and the order in which the block is zigzag scanned. Is low. The coefficient encoding unit 102 scans the coefficients in the order of number 1 to number 63 in FIG. 4 to check whether they are zero or not. However, since the coefficient of number 0 is a DC coefficient, zigzag scanning is not performed.
[0025]
Process S12 in FIG. 3 is a preparatory process for starting a zigzag scan. A counter for specifying a coefficient is initialized to 1, and a counter for counting the number of zero coefficients is initialized to 0.
[0026]
The process S13 is a determination process for checking whether the zigzag scan of the block has been completed. This checks whether the counter specifying the coefficient is less than 64. If the result is Yes, the process proceeds to step S14. If the result is No, the process proceeds to step S19.
[0027]
The process S14 is a determination process for checking whether the coefficient designated by the counter is 0. If the result is Yes, the coefficient is 0 and the process proceeds to step S21. If the result is No, the process proceeds to step S15.
[0028]
Process S15 is a determination process for checking whether the spatial frequency of the coefficient is on the low frequency side as compared with the threshold value specified by the spatial threshold value specifying means 103. In process S15, the counter value specifying the position of the coefficient is compared with the threshold value specified by the spatial threshold value specifying means 103, and it is checked whether the counter value is less than the threshold value. If the result is Yes, the process proceeds to step S16, and if the result is No, the process proceeds to step S18. The spatial threshold designating unit 103 changes the threshold based on the MCU counter value. When the MCU counter designates the vicinity of the center of the image, the threshold value is set to 64 in order to emphasize the image quality. When the threshold value is designated as 64, all non-zero coefficients are determined to be less than the threshold value, and are encoded and output in step S16. On the other hand, when the MCU counter designates the periphery of the image, the threshold is set to 15. As a result, when there is a non-zero coefficient between 1 and 14 in the coefficient counter, it is guaranteed that the coefficient counter is not replaced with zero and is encoded in step S16.
[0029]
The reason why the low frequency side coefficient is excluded from the code amount control target even in the peripheral MCU that does not place importance on the image quality is that if the low frequency side coefficient is zero-substituted, the influence on the decoded image is large. On the other hand, the coefficient on the high frequency side has a relatively small effect even if zero substitution is performed, and therefore, the coefficient is deleted when executing the code amount control. In addition, the threshold value is changed according to the position of the MCU image notified by the spatial threshold value designating means 103 for the image data block located near the center of the image to invalidate the code amount control and guarantee the image quality. This is because the code amount control is enabled for the blocks located in the periphery of the image while allowing the image quality to deteriorate.
[0030]
In step S16, the coefficient encoding unit 102 executes encoding and outputs a code, and decrements the number of codes output from the allowable number storage means 105. Encoding is performed as follows. While the number of zero coefficients is 16 or more, the ZRL code is output and the operation of subtracting 16 from the number of zero coefficients is repeated. ZRL is Zero Run Length, which exists as one of the code symbols in the two-dimensional Huffman code table. If the number of zero coefficients is less than 16, the non-zero coefficient to be encoded is decomposed into the level value to which the coefficient belongs and the remaining bits within that level, and the number of zero coefficients and the non-zero coefficient The pair with the level value is two-dimensionally Huffman encoded, and then the remaining bits are attached. Both the ZRL code and the two-dimensional Huffman code decrement the allowable number storage means 105 for each code output. After outputting the code, a zero coefficient counter is further initialized to 0 in step S16. Thereafter, the process proceeds to S17.
[0031]
In the process S17, the counter for designating the coefficient position is incremented and the process returns to the determination process S13.
[0032]
When the zigzag scan of all the AC coefficients is completed, the process proceeds from the determination process S13 to the process S19. Process S19 is a determination process for checking whether the number of zero coefficients is zero. If the determination result is Yes, the encoding process for the block ends. If the result is No, an EOB code is output and the allowable number storage means 105 is decremented by one. The EOB is an end of block, and is present in the two-dimensional Huffman code table as one of the code symbols.
[0033]
The coefficient encoding unit 102 outputs the two-dimensional Huffman code when encoding a non-zero AC coefficient, when outputting ZRL, or when outputting EOB. Since an AC coefficient rarely appears at the end of a block, an EOB code is output at the end of the block in many cases. In the present invention, the fact that the correlation between the number of code outputs of the coefficient coding unit and the code amount is high is used, the number of code outputs is measured, and the number of code outputs is controlled so that this is within the allowable number. Is. In the present embodiment, ZRL and EOB are included when counting the number of code outputs, but an image encoding device that counts without including these may be created.
[0034]
In the present invention, the code amount control works and the actual reduction operation is executed in step S18. Process S18 is a determination process for checking whether the value stored in the allowable number storage means 105 is positive. If a determination result is Yes, it will progress to process S16 and will output a code | symbol. In No, it progresses to process S21. Process S21 is a process of incrementing the counter for counting the number of zero-valued coefficients by one, and then proceeds to process S17. When the process proceeds from the process S18 to the process S21, the coefficient that was not zero at the time of quantization is processed as zero.
[0035]
Next, a method for realizing the code amount control while changing the image quality according to the position in the image will be described. FIG. 5 shows an image 110 and MCUs constituting the image 110. The size of the MCU is 16 pixels horizontally and 16 pixels vertically, and includes four luminance component blocks and two color difference component blocks. Six MCUs are arranged in the horizontal direction of the image 110, and seven MCUs are arranged in the vertical direction. If the MCU is specified by a set of a horizontal index and a vertical index, the MCU of (0, 0) is the MCU 111, and the MCU of (5, 0) is the MCU 116. By using a vertical index × 6 + horizontal index as the MCU counter, the position of the MCU in the image can be specified by one counter value. In FIG. 5, the MCU 118 and the like located near the center of the image 110 are drawn by solid lines, and the MCU 111 to MCU 117 and the like located around the image 110 are indicated by dotted lines. The importance is low because it is located, and the MCU indicated by the solid line is high in importance because it is located in the center, indicating that there is a difference in the priority of image quality.
[0036]
The encoding apparatus of the present invention controls the spatial code amount distribution by the operation of the spatial allowable number increase means 104, and how many values to be stored in the allowable number storage means 105 are increased in step S03 in FIG. It must be designed according to the target image size and code amount.
[0037]
First, the total allowable number is determined in advance how many two-dimensional Huffman codes can be output for the entire image. Since there is a strong correlation between the code amount of the image and the number of encoded 2D Huffman codes in the encoding result, a quantization table that can be encoded within the target code amount for the target standard image is obtained by trial and error. At this time, the number of two-dimensional Huffman codes output in the entire image is measured, and this is defined as Ni.
[0038]
Next, the obtained Ni is divided by the total number of MCUs, and the number of two-dimensional Huffman codes per MCU is obtained, and this is defined as Nm. Here, 40 two-dimensional Huffman code outputs Nm per MCU are set. This is a number obtained as a result of selecting a quantization table by setting the target code amount per MCU to about 17 bits.
[0039]
In order to distribute the code amount according to the spatial position in the image of the block, the spatial allowable number increase means 104 stores the allowable number increase value according to the position in the MCU image, and the position in the MCU image. Is input, the allowable number increase value is read and added to the stored value of the allowable number storage means 105 to be updated.
[0040]
FIG. 6 shows an increase value of the allowable number for each MCU. FIG. 6 shows an increase value of the allowable number at the position of each MCU constituting the image 110, and the spatial allowable number increase means 104 stores this value in a table.
[0041]
First, encoding is performed on the MCU 111, and the position of the MCU is (0, 0), and the value of the MCU counter representing this position is notified to the spatial allowable number increasing means 104 in step S03 in FIG. The number increase means 104 refers to the position of (0,0) in the table and increases the stored value of the allowable number storage means 105 by 40. This means that since Nm is 40, one MCU is allocated, and code amount control does not work until 40 codes are output. The coefficient encoding unit 102 encodes the coefficient of the MCU 111 and decreases the value stored in the allowable number storage unit 105 by 1 each time a two-dimensional Huffman code is output.
[0042]
Since the MCU 111 is located in the vicinity, and the spatial threshold value specifying unit 103 sets a small threshold value, when the number of codes to be output is large, the stored value of the allowable number storage unit 105 is not positive, and the coefficient encoding is performed. The unit 102 regards the high frequency side coefficient equal to or higher than the threshold as zero and limits the code output.
[0043]
When encoding of the MCU 111 is completed, the spatial allowable number increasing means 104 increases the stored value of the allowable number storage means 105 by 240 before encoding the MCU 112 at the position (0, 1) next. This is an allowable number of 6 MCUs, that is, 6Nm, and is an allocation for encoding from the MCU 112 at the (0,1) position to the MCU 117 at the (1,0) position. Thereafter, the allowable number increase value at the time of encoding from MCU 113 to MCU 117 is set to zero. These increased values are stored in a table in the spatial allowable number increasing means 104.
[0044]
If encoding is performed in order from the MCU 112, since the number of allowable times is large in the previously encoded MCU, all coefficients are encoded without being limited in code output. For this reason, if the MCU that is encoded earlier outputs many coefficients, the value of the allowable number storage means 105 is not positive in the MCU that is encoded later, and the output of the high frequency side coefficient is likely to be limited. In particular, output restriction is likely to occur in the MCU 117 at the position (1,0) and the MCU 116 at the position (0,6), so that the image quality around the image is degraded. On the other hand, in the MCU 112 at the position (0, 1), the subsequent MCU 113, and the like, there are abundant values in the allowable number storage means 105, so there is no code output limitation, and the image quality is quantized by the quantization table. The image quality will be as shown.
[0045]
Similarly, as shown in FIG. 6, j is an integer from 1 to 5, and the allowable number increase value for the MCU at the position (1, j) is set to 240 including the subsequent MCUs, and the allowable values for other MCUs are set. Set the increment value to 0. As a result, the MCU at the position (1, j) always guarantees good image quality, and the image quality of the MCU at the position (0, j + 1) and the position (5, j) tends to deteriorate. Similarly, when j is 6, before encoding the MCU located at the position (1,6) as shown in FIG. 6, the spatial allowable number increase means 104 increases the stored value of the allowable number storage means 105 by 200. . This is an allowable number of 5 or 5Nm.
[0046]
For the MCU located near the center of the image, for example, the MCU 118 in FIG. 5, the code output by the coefficient encoding unit 102 is not limited in order to avoid image quality degradation. This is realized by changing the threshold specified by the spatial threshold specifying means 103 in accordance with the position of the image data in the image. Since the 8 × 8 coefficient block is zigzag scanned in the order shown in FIG. 4, the spatial frequency of the coefficient can be divided into bands in which the sum of the horizontal and vertical indexes is constant. Therefore, the spatial threshold designation unit 103 notifies the threshold in the zigzag scan order of FIG. 4, and the coefficient encoding unit 102 thereby determines the spatial frequency of the coefficient.
[0047]
Therefore, in the peripheral MCU indicated by the dotted line in FIG. 5, the spatial threshold value specifying means 103 notifies 15 as the threshold value. The coefficient whose spatial frequency is less than 15 is the 14 AC coefficients numbered 1 to 14 in FIG. 4. If the sum of the horizontal and vertical indices of the coefficient is less than 5, the low frequency side It will be judged. When the coefficient encoding unit 102 finds a non-zero coefficient in the determination process S14 of FIG. 3, the coefficient encoding unit 102 checks whether the counter representing the position of the coefficient is less than the threshold value in the determination process S15, and outputs a code if the result is Yes. If so, it is checked in the determination process S18 whether the value of the allowable number storage means 105 is positive. If the result is Yes, the code is output in the process S17, and if it is No, the output of the two-dimensional Huffman code of the coefficient is output. Restrict.
[0048]
On the other hand, when the MCU indicated by the solid line in FIG. 5 is processed, the threshold value is set to 64 when the value of the MCU counter is notified to the spatial threshold value specifying means 103 in step S03 of FIG. As a result, all AC coefficients are determined to be less than the threshold value in the determination process S15, and the coefficient encoding unit 102 does not perform code output restriction. That is, all the non-zero coefficients found in the determination process S14 are encoded in the process S16, and the stored value in the allowable number storage unit 105 is decremented each time a code is output. As a result, it is possible to obtain an image encoding device that can control the image quality and code amount of the peripheral portion of the image while maintaining the image quality near the center of the image.
[0049]
In this embodiment, the spatial allowable number increase means 104 stores the increase value of the allowable number in the table and updates the value of the allowable number storage means 105 with reference to the table. It can also be used. For example, an increase value may be obtained by calculation based on the MCU address, and the value in the allowable number storage means 105 may be updated. In addition, this program reflects the past code generation situation, predicts an allowable increase in the number of times required in the next MCU, and adaptively disperses image quality degradation due to insufficient allowable numbers to neighboring MCUs. It is also possible to execute various controls.
[0050]
Furthermore, in the present embodiment, the spatial allowable number increasing means 104 is operated so that the value of the allowable number storage means 105 is updated before encoding the MCU in units of MCUs. It is also possible to input and operate the spatial allowable number increasing means 104 more finely in units of blocks.
[0051]
【The invention's effect】
According to the image coding apparatus of the present invention, the following effects can be obtained. The image quality can be controlled according to the position in the image while executing the code amount control so as to be within a certain code amount. In particular, image quality degradation occurs in the peripheral portion of the image, and a constant image quality is guaranteed in the central portion of the image, and information in the central portion that is important in the image is not degraded. In addition, the output code can be decoded by an international standard JPEG decoder, no dedicated decoder is required, and various popular JPEG decoders can be used, increasing the degree of freedom of the system. In addition, the production cost of the decoder can be reduced. Furthermore, the output code can be reused for various purposes without format conversion.
[Brief description of the drawings]
FIG. 1 is a block diagram of an image encoding apparatus according to the present invention.
FIG. 2 is a flowchart showing the operation of the image coding apparatus according to the present invention.
FIG. 3 is a flowchart showing an operation of coding a block by a coefficient coding unit according to the present invention.
FIG. 4 is a diagram showing the order of zigzag scanning of an 8 × 8 coefficient block.
FIG. 5 is a diagram showing MCUs located in a peripheral part and MCUs located in a central part among MCUs constituting an image;
FIG. 6 is a diagram showing an increase value of the allowable number that is increased by the spatial allowable number increase means before encoding each MCU in correspondence with the position of the MCU.
FIG. 7 is a block diagram of a conventional image encoding device.
[Explanation of symbols]
101: Transform coding unit
102: Coefficient encoding unit
103: Spatial threshold designation means
104: Means for increasing the allowable number of times of space
105: Permissible number storage means
110: Image
111-118: MCU
201: DCT calculation unit
202: Quantization unit
203: VLC section
204: VLC control circuit
205: Rate calculation circuit

Claims (1)

In an image coding apparatus comprising transform coding means for discrete cosine transform and quantizing image data for each block and outputting coefficients, and coefficient coding means for coding coefficients output by the transform coding means,
An allowable number storage means for storing the allowable number of code outputs of the coefficient encoding means;
Spatial allowable number increase means for increasing the allowable number of times stored in the allowable number storage means according to the position of the image data in the image;
Spatial threshold value specifying means for specifying a threshold value of a coefficient according to the position of the image data in the image;
With
The coefficient encoding unit compares a counter value indicating the position of the coefficient output by the transform encoding unit in the zigzag scan order with a threshold value specified by the spatial threshold value specifying unit, and when the counter value is lower than the threshold value Encodes the coefficient and outputs the code. If the allowable number stored in the allowable number storage means is positive, the coefficient is encoded and the code is output. If the allowable number is not positive, the coefficient is treated as zero. Does not output the sign,
Further, when the coefficient encoding means outputs a code, the image encoding apparatus controls to reduce the allowable number of times stored in the allowable number storage means by the number of the outputted codes.

Images