JPH0678287A - Coding method and device - Google Patents

Abstract

PURPOSE:To prevent generation of a useless code by revising a deciding criterion to discriminate coding processing based on transition of coded data quantity being generated for one frame at least. CONSTITUTION:A processing selection circuit 27 references block data inputted from a signal line 103, a quantization step (g) from a signal line 101 and preceding frame data stored in a preceding frame memory 28 to discriminate the processing system. When it is discriminated that an inter-frame difference coding method is proper, block data in the preceding frame memory 28 and at the same location as an input block are read and outputted to a signal line 114. When it is discriminated that an inter-frame difference coding method with motion compensation is proper, block data of best matching in the preceding frame memory 28 are read via a signal line 106 and outputted to the signal line 114. When it is discriminated that an in-frame coding method is proper, zero data are generated and outputted to the signal line 114.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding method and apparatus, and more particularly to a moving picture coding method and apparatus.

[0002]

2. Description of the Related Art In the case of a moving image, for example, about 30 frame images are sequentially displayed in one second, and as a result, a person recognizes it as a moving image. Therefore, it is important to encode each frame image efficiently and without impairing the quality.

Generally, the following three are known as frame encoding.

1. Inter-frame differential encoding method, 2. 2. Inter-frame differential encoding method with motion compensation, 3. Intra-frame coding method These are appropriately used to perform video coding as a whole.

Each coding method will be described below.

1. Interframe Difference Coding Method An image for one frame is composed of, for example, n × n pixel blocks (hereinafter, simply referred to as blocks). In this encoding method, the difference between the information of a certain block in the frame to be encoded and the block information of the same position in the previous frame is calculated, and the difference is encoded. In this inter-frame differential encoding method, the higher the correlation between the frames, in other words, the smaller the change in the motion of the image, the closer the inter-block difference becomes to zero, so the encoding efficiency is high. Become. For the difference data block, Discreate Cosine Trans
form) processing is performed, and the obtained transform coefficient data is quantized and Huffman coding is performed. Then, when all the data become zero after the quantization, it means that the quantized image is the same as the same block in the previous frame, and therefore the code data is not transmitted.

2. Inter-frame differential encoding method with motion compensation In this encoding method, the block to be encoded,
Each block is matched with a plurality of neighboring blocks centered on the block at the same position on the previous frame, and the most similar block is selected. Then, the difference between the block to be encoded and the selected block is calculated. Then, the difference data thus obtained is subjected to discrete cosine transform and quantized. Then, Huffman coding is applied to the quantized data.

3. Intra-frame Encoding Method This encoding method is a method in which the entire frame to be encoded is directly subjected to discrete cosine transform, the obtained transform coefficient data is quantized, and then encoded. In the two encoding methods described above, the encoding efficiency can be increased when the correlation with the previous frame is high, but when the inter-frame correlation is small, in other words, when the image is switched such as scene change, the dynamic The range becomes large, and the amount of information of encoded data is increased. This intraframe coding method is effective when the correlation between the frame of interest and the previous frame is small.

[0009] Incidentally, the MP of the International Standardization Committee (ISO)
"DCT (Discrete Cosine Transform) + Motion Compensation (Moti)
on Compensation) ", the data coded by this intra-frame coding method is written in units of 15 frames (when displaying 30 frames per second, 0.5 second intervals).
The data is transferred as RA frame (intra frame) data. That is, if a certain number (or time) of continuous encoding is performed based on the correlation with the previous frame,
It deals with the fact that the quantization error becomes large and it looks unnatural to humans.

By the way, generally, in the case of generating an INTRA frame with such a fixed cycle, there is no problem if the video dynamically changes when it happens to match the cycle, but it is matched with the previous frame. If it occurs during the encoding by, an unnatural video is displayed momentarily.

Therefore, for example, it is conceivable to switch these three coding methods at appropriate times.

Description will be made with reference to the flowchart of FIG.

The overflow signal (signal 304 in FIG. 19) in the following description is a signal indicating that the amount of data in the buffer memory for temporarily storing encoded data has reached a predetermined size, and to some extent. It is a signal with a margin of.

First, in step S101, it is determined whether or not an overflow condition exists. If it is in the overflow state, it is decided to perform the inter-frame coding process on the pixel block of interest.

If it is determined that the overflow state has not been reached, the process proceeds to step S102, and it is determined whether or not the inter-frame differential encoding method with motion compensation is performed.
Specifically, as shown in FIG. 17, the difference value of the target block is subjected to the motion compensation inter-frame differential encoding process and the difference value of the target block is subjected to the inter-frame differential encoding process without motion compensation. It is determined whether or not the determined point is in the "MC" area shown. Although it may be possible to determine whether the line passes through the origin and is above or below a straight line 1700 having a slope of m = 1, vector information for the target block is also generated in the inter-frame differential encoding processing by motion compensation. A straight line 1701 shown in the figure is obtained by adding the data amount of this vector information.

If it is determined in step S102 that the target block is within the "MC" area, it is determined to perform the frame difference encoding process with motion compensation.

If it is within the "inter" area, the inter-frame differential encoding process with motion compensation is not performed, and the process proceeds to step S103.

In step S103, it is determined whether to perform the intra-frame coding process or the inter-frame coding process. Specifically, the difference variance value of inter-frame data when motion compensation is performed on the block of interest,
A difference variance value when encoded by a process without motion compensation is obtained, and if the target block is in the "INTRA" area in FIG. 18, the intraframe encoding process is selected, and "inte
If there is an area in r ″, the interframe differential encoding process is selected.

In the process selection described above,
The selection is performed using the statistics of the luminance data 16 × 16 pixel block.

The configuration of the encoder will be described with reference to FIG.

In FIG. 19, the block data input from the input line 301 is the subtractor 30 and the mode determiner 3
Input to 7. On the other hand, the signal line 304 is an input signal line for the overflow signal of the output code buffer and is input to the mode determiner 37 and the mask 32.

The mode determiner 37 refers to the block data input from the signal line 303 and the previous frame data stored in the previous frame memory 38, and determines the processing method according to the above method. If it is determined that the interframe differential encoding method is suitable, the data of the block at the same position as the input block on the previous frame memory 38 is read and output to the signal line 308. On the other hand, when it is determined that the inter-frame differential encoding method with motion compensation is suitable, the best matching block data on the previous frame memory 38 is read out via 307 and output to the signal line 308.
Furthermore, when it is determined that the intraframe coding method is suitable, zero data is generated and is output from the signal line 308. At the same time, vector data representing the relative position between the best matching block obtained by the motion compensation and the coding block is output from the signal line 312.

The subtractor 30 takes the difference between the block data from the signal line 310 and the input block data from the signal line 302. This difference data is DCT (Discrete Cos).
ineTransform: Discrete Cosine Transform) circuit 31 and input to mask section 32.

On the other hand, when the signal from the input signal line 304 indicates an overflow state, the mode determiner 37
Uniquely selects interframe differential encoding. Further, in this case, the mask unit 32 masks all difference data to zero. The DCT transform coefficient data from the mask unit 32 is quantized by the quantizer 33 and input to the encoder 34 and the inverse quantizer 35.

The encoder 34 selects the mode signal 31 from the mode determiner 37 for the quantized transform coefficient data.
The Huffman code is assigned with reference to 1 and output from the signal line 313. The inverse quantizer 35 reproduces the same frequency data as that sent to the external decoder (not shown) by inverse quantization. The reproduced frequency data is used as the inverse DCT circuit 36.
Is again converted into a differential signal by the adder 39 and the signal from the signal line 309 is added by the adder 39 to reproduce the same image as that transmitted. This is again stored in the previous frame memory 38.

[0026]

In the processing described above, the quantization step tends to increase when the information amount of the image is very large with respect to the transmission rate. Therefore, the data that is actually quantized and encoded and transmitted is
There is a drawback that it is limited to those that have a sufficiently large amount of information before quantization.

Further, which encoding process is to be performed is fixedly determined. As a result, improper coding processing may be performed even on a block in which no data remains after the inter-frame difference is quantized, that is, a block that does not need to send a code in the current quantization step. Therefore, there is a drawback that unnecessary codes are transmitted.

When the encoding process is performed at a constant bit rate, the average code amount assigned to one block is extremely small. For this reason, for the block with a small difference between the previous frame and the current frame, the method of using the data of the previous frame as it is and not transmitting the code as the difference at all is adopted, and the code is allocated to the block with a large difference. Increasing is done.

However, when the correlation between two adjacent frames is small over the entire surface such as a scene change, all blocks require codes, and the codes generated in individual blocks become large. At this time, in the above encoding method, a large amount of codes are generated by encoding one block, and many other blocks are not encoded.

[0030]

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and it is for selecting an encoding process based on the transition of the encoded data amount which is occurring for at least one frame. It is an object of the present invention to provide an encoding method and apparatus that can change the criterion and prevent the generation of unnecessary codes.

In order to solve this problem, the encoding method of the present invention comprises the following steps. That is, a coding method that codes individual pixel blocks in units of pixel blocks of a predetermined size that form a frame image, and is based on the correlation between the pixel block in the frame of interest and the corresponding block in the previous frame. Then, based on the determination process for determining whether the intra-frame coding process or the inter-frame differential coding process is performed on the pixel block of interest and the amount of coded data in the past, the determination criteria in the determination process are set. And an adjusting step for adjusting.

The coding apparatus of the present invention has the following configuration. That is, an encoding device that encodes individual pixel blocks in units of pixel blocks of a predetermined size that form a frame image, is based on the correlation between the pixel block in the frame of interest and the corresponding block in the previous frame. And a determination means for determining whether the intra-frame encoding process or the inter-frame differential encoding process is performed on the pixel block of interest, and the determination criterion in the determination process based on the encoded data amount in the past. And adjusting means for adjusting.

According to another aspect of the present invention, the masking size after the orthogonal transform is changed based on the transition of the coded data amount that is occurring for at least one frame, so that the code of one block can be changed. It is an object of the present invention to provide an encoding method and apparatus that can suppress excessive occurrence.

In order to solve this problem, the encoding method of the present invention comprises the following steps. That is, a pixel block of a predetermined size that constitutes a frame image is selected in an encoding method in which one of the encoding processes selected based on the correlation with the corresponding block of the previous frame is selected and encoded. The orthogonal conversion process of orthogonally converting the target block to which the difference information is added based on the encoding process and the masking ratio of the AC component in the orthogonally converted block data based on the amount of encoded data up to the previous time are adjusted. And an adjusting process.

The coding apparatus of the present invention has the following configuration. That is, a pixel block of a predetermined size that forms a frame image is selected by an encoding device that selects and encodes one of the encoding processes that is selected based on the correlation with the corresponding block of the previous frame. The orthogonal conversion means for orthogonally converting the target block to which the difference information is added based on the encoding process described above, and the masking ratio of the AC component in the orthogonally converted block data based on the amount of encoded data up to now are adjusted. And adjusting means for performing the adjustment.

Another aspect of the present invention is to provide a coding method and apparatus that enable more efficient coding processing by adding new coding processing in addition to normal coding processing. It is the one we are trying to provide.

In order to solve this problem, the encoding method of the present invention comprises the following steps. That is, a pixel block of a predetermined size that constitutes a frame image is selected in an encoding method in which one of the encoding processes selected based on the correlation with the corresponding block of the previous frame is selected and encoded. The first conversion step of converting the pixel block to which the difference information based on the encoding process described above is added in the frequency space and the type information unit included in the raw pixel block are averaged, and the entire pixel block is averaged value data. A second conversion process for converting into the first conversion process and a selection process for adaptively selecting the first conversion process and the second conversion process.

The coding apparatus of the present invention has the following configuration. That is, a pixel block of a predetermined size that forms a frame image is selected by an encoding device that selects and encodes one of the encoding processes that is selected based on the correlation with the corresponding block of the previous frame. First conversion means for converting the pixel block to which the difference information based on the encoding processing is added in the frequency space, and averaging in the type information unit included in the raw pixel block, and the entire pixel block as average value data. A second converting means for converting to the first converting means, and a selecting means for adaptively selecting the first converting means and the second converting means.

[0039]

In the process or configuration of the present invention, the criterion for determining the coding process to be selected for the block of interest in the frame of interest is adjusted based on the amount of coding up to now. .

[0040]

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

[First Embodiment] FIG. 1 is a block diagram of an encoder according to an embodiment of the present invention. First, the encoder of the present embodiment will be described with reference to FIG.

In FIG. 1, the main difference from FIG. 19 described above is that a mode selection unit 37 of FIG. 19 is replaced with a process selection circuit 27 and several signal lines are added. .

The other structure is similar to that of FIG.

In FIG. 1, the block data input from the input line 103 is the processing selection circuit 27 and the subtractor 20.
Entered in. On the other hand, the signal line 102 is an input signal line for an overflow signal of an output code buffer (not shown) and is input to the process selection circuit 27. The meaning of the overflow signal is as described above.

The processing selection circuit 27 refers to the block data input from the signal line 103, the value of the quantization step g from the signal line 101, and the previous frame data stored in the previous frame memory 28, and executes the processing method described later. Determine the processing method. If it is determined that the inter-frame differential encoding method is suitable, the block data on the previous frame memory 28 at the same position as the input block is read and output to the signal line 114.

On the other hand, when it is determined that the inter-frame differential encoding method with motion compensation is suitable, the previous frame memory 28
The block data of the above best matching (having the closest configuration) is read out via 106 and output to the signal line 114. At the same time, vector data representing the relative position between the best matching block obtained by motion compensation and the block to be encoded is output from the signal line 116. Further, when it is determined that the intraframe coding method is suitable, zero data is generated and is output from the signal line 114. The block data on the signal line 114 is output to the subtractor 20 and the adder 29.

The subtractor 20 takes the difference between the block data from the signal line 114 and the input block data from the signal line 103. This difference data is DCT (Discrete Cos).
ineTransform (discrete cosine transform) circuit 21 and supplied to the mask unit 22.

On the other hand, when the overflow signal from the input signal line 102 indicates the overflow state of the code buffer, the process selection circuit 27 uniquely selects the interframe differential encoding process. The DCT transform coefficient data from the mask unit 22 is quantized by the quantizer 23, and the encoder 24
And to the inverse quantizer 25.

The encoder 24 selects the selection mode signal 11 from the process selection circuit 27 for the quantized transform coefficient data.
5, a Huffman code is assigned and output from the signal line 207 to the output code buffer memory. Inverse quantizer 25
Then, the same frequency data as that sent to the external decoder (not shown) is reproduced by the inverse quantization. The reproduced frequency data is converted into a differential signal again by the inverse DCT circuit 26. The adder 29 adds the output data of the inverse DCT circuit 26 and the block without front frame obtained from the signal line 114, reproduces the same image as that transmitted, and stores it in the previous frame memory 238.

Next, FIG. 2 shows a detailed configuration of the process selection circuit 27 in FIG.

In FIG. 2, a signal line 101 is a signal line for the quantization step g, and a signal line 102 is a signal line for inputting an overflow state. The image block data is input from the signal line 103 to the intra-frame correlation calculation circuit 3, the inter-frame correlation calculation and motion compensation circuit 2.

The meaning of the signal at the quantization step g101 is
It is a value indicating how many the transform coefficient of the orthogonal transform is divided.
Briefly, it is a signal that indicates how many divisions the range width for quantization should be divided into.The smaller the value, the more the quantization error is suppressed, and conversion is performed with high quality. Data tends to be redundant. The quantization step g signal is determined by the amount of accumulated coded data generated so far. Of course,
When the amount of coded data in the past is large, the value of the quantization step g becomes large, and conversely, when the amount of coded data is small, the value becomes small. However, the generation principle of the quantization step g will be described later.

Then, the mask circuit 22 changes the masking range of the high frequency component of the orthogonal transformation based on the quantization step g signal. Specifically, the larger the value of the quantization step g signal 102, the larger the ratio of masking the AC component. Details of this masking process will also be described later. Further, the quantizer 23 quantizes the data from the mask circuit 22 in a quantization step corresponding to the quantization step g signal 102.

In FIG. 2, the interframe correlation calculation and motion compensation circuit 2 reads the previous frame block data from the previous frame memory 28 via the signal line 106. Then, the absolute value of the difference between the pixels in the respective blocks is calculated, and the average thereof is calculated to calculate the inter-frame difference absolute value average Dif. Then, the calculated Dif value is output to the signal line 109. The block data on the previous frame at the same position as the position of the block of interest is output to the signal line 112.

In addition to the above-described processing, the inter-frame correlation calculation and motion compensation circuit 2 searches for the best matching block with the smallest absolute difference value in the vicinity of the target block position on the previous frame, and retrieves this signal line. In addition to the output from 113, the motion compensation frame difference absolute value average MCDif between the obtained block and the input block from the signal line 103 is calculated and output from the signal line 110. The calculation method of Dif and MCDif is expressed by the following equation E-1,
It is shown in E-2.

The inter-frame correlation calculation and motion compensation circuit 2 calculates the inter-frame root mean square value VarDif with motion compensation according to the equation E-3, and outputs it from the signal line 111.

In the inter-frame correlation calculation and the motion compensation in the motion compensation circuit 2, the matching value with the block shifted by one pixel unit within the distance range of about 7 pixels vertically and horizontally from the block at the same position on the previous frame, E- shown below
The best matching block is calculated by the formula 4 and the block giving the smallest one is calculated.

The intra-frame correlation calculator 3 uses the signal line 103
The variance value VarOr is obtained from the block data input via the E-5 equation, and this is output to the selector 1 from the signal line 107. At the same time, a dummy block of all zero data is generated, which is also signal line 108. Output to selector 1.

(E-1) Dif = 1 / N ² ΣΣ│B _ij -PB _ij │ where N: block size, B _ij : block of interest for coding, PB _ij : block on previous frame.

(E-2) MCDif = 1 / N ² ΣΣ│B _ij -DB _{i + m, j +} n│ where DB is the best matching block by motion compensation. m = ... -3, -2, -1, 1, 2, 3, ...

(E-3) VarDif = 1 / N ² ΣΣ (B _ij −DB _{i + m, j + n} ) ² (Equation E-4) Matching value = ΣΣ (B _ij −MCB _ij ) ² However, MCB: A neighborhood block on the previous frame.

(E-5) VarOr = 1 / N ² ΣΣ (B _ij ) ^2- {1 / N ² ΣΣ (B _ij )} ² Then, the selector 1 receives the input signal Dif 109,
MCDif110, VarDif111, VarOr1
07, g value 101, and overflow information 102, any block data of the signal line 112, the signal line 113, and the signal line 108 is selected and output to the signal line 114.

The method of determining the signal line to be selected by the selector 1 will be described below with reference to the flow chart of FIG.

In FIG. 3, first, in step S1 which is the judgment A, it is checked whether or not the code generation amount so far is large and the overflow state is present. If there is an overflow, the signal line 112 is used to select the interframe differential encoding method.
The same position block data on the previous frame of is selected.

On the other hand, if there is no overflow in step S1, the process proceeds to decision B in step S2, and it is determined whether or not the interframe differential coding method with motion compensation is performed.
Specifically, Dif and MC from the signal lines 109 and 110
The determination is made by Dif. If the relationship between MCDif when motion compensation is performed according to the characteristics shown in FIG. 4 and Dif when motion compensation is not performed is within the MC range, motion compensation with inter-frame differential encoding should be selected. The best matching block data 113 according to is selected. The threshold line f (g) for the determination here changes depending on the g value from the signal line 101.

If it is determined that the motion compensation is not performed within the Inter range, the inter-frame differential encoding method with motion compensation is not selected and the determination C in step S3 is performed.
Proceed to. In step S3, according to VarDif and VarOr from the signal lines 111 and 107, it is determined whether the interframe coding method or the intraframe coding method is used by the boundary line shown in FIG. Be done. This boundary line F (g) is also a function of g as shown in the figure, and changes according to the g value.

If it is within the INTRA range, the intra-frame encoding process will be performed, so "0" block data on the signal line 108 is selected. As a result, the block data of interest does not change at all in the subtractor 20 and remains DC
It is output at T21. If it is within the “Inter” range, it is determined to be interframe coding, and the same-position block on the previous frame from the signal line 112 is selected.

Through the above processing, the selected block data is output from the signal line 114, and the selected mode is output from the signal line 115.

Next, the quantization step signal generation circuit for generating the quantization step signal g102 in the embodiment will be described.

FIG. 20 shows a concrete example around the quantization step signal generation circuit. Note that the value of the quantization step g signal in this embodiment changes in units of 16 lines since this apparatus has a size of 16 × 16 dots for one scan, or more specifically, one block has a size of 16 × 16 dots (hereinafter, this unit is Block line). In other words, the quantization step g signal does not change in one block line. The reason will be explained later.

In FIG. 20, the data including the code data for the block data output from the encoder 24 shown in FIG. 1 is transferred to the output code buffer and also output to the code length detection circuit 400. The code length detection circuit detects the effective number of bits of the code data of the block of interest and outputs the value as the signal 400a to the subtractor 401.
Output to. The subtractor 401 takes a difference from a preset standard value, and outputs a signal indicating how large a code is generated or how small a code is generated with respect to the standard value.

On the other hand, "0" is set in the register 402 as an initial value at the start of encoding one frame, and the stored contents are supplied to the adder 403. That is, it is reset in synchronization with a vertical synchronizing signal (not shown) of one frame to be encoded. The adder 403 adds the value from the register 402 and the value from the subtractor 401, and feeds back the result to the register 402.
Therefore, each time the code data for the block data is generated, the operation of this circuit causes the register 40 to operate.
In 2 is stored and held a value indicating how large or small the encoded data is generated with respect to the standard number of code bits.

In the quantization step signal generator 404,
There is a counter (not shown) for counting horizontal synchronization signals, and every time 16 horizontal synchronization signals are detected by the counter, the content held in the register 402 is read,
A 3-bit quantization step g signal is generated based on the value and its output is held. Since it is a 3-bit signal line, a total of 8 kinds of signals 0 to 7 can be generated.
In this embodiment, signals 0 to 5 are generated.

Incidentally, when the value stored in the register 402 after the encoding of one block line is large,
That means that the encoded data amount in the previous block line was larger than the standard encoded data amount.
Therefore, in this case, the quantization step g signal is increased to suppress the generation of encoded data. An overflow signal generation circuit 405 generates the overflow signal 102 according to the value of the register 402.

As a result of the above, as shown in FIGS. 4 and 5, the border line for deciding which of the inter-frame differential encoding process, the inter-frame differential encoding process with motion compensation, and the intra-frame encoding process is to be adopted is , It will vary in units of one block line depending on the size of the quantization step. That is, the criterion for deciding which encoding process should be performed varies depending on the size of the quantization step. This result also affects the encoder 24 and can contribute to the prevention of unnecessary code generation and transmission.

Whether the process selection method and intra-frame coding method for deciding whether or not the inter-frame differential coding method with motion compensation according to the present embodiment is performed, or the inter-frame differential coding method is selected. The process selection method that determines
The determination criteria are not limited to those shown in FIGS. 4 and 5 described above, and an appropriate determination method may be selected in a timely manner. 6 and 7 show other determination methods that can be implemented instead of the determination methods shown in FIGS. 4 and 5 described above.

FIG. 6 is a diagram showing another processing selection method for determining whether or not the inter-frame differential coding method with motion compensation of this embodiment is performed, and FIG. 7 is the intra-frame coding method of this embodiment. It is a figure showing the other process selection method which determines whether to select, or the inter-frame difference encoding method. Figure 6
If the range includes the motion compensation shown in (1), the inter-frame differential encoding method with motion compensation is executed. If there is no motion compensation, the process proceeds to step 3 and the inter-frame differential encoding method is performed according to the determination standard shown in FIG. Here, the intra-frame coding method is executed when the range is within the frame, and the inter-frame differential coding method is selected and executed when the range is between the frames.

The code amount control in this embodiment in the above processing will be described below.

At the time of data transmission, it is necessary to control the amount of coded data generated to match the transmission rate. In this embodiment, this transmission speed matching is executed by the method described below.

That is, the generated code is temporarily stored in the output code buffer having a code amount corresponding to one frame, and is sequentially transmitted from there at a speed corresponding to the transmission rate. The encoder monitors the sufficiency of the code in the buffer, and when the code is generated too much, the quantization step is increased to suppress the code generation. On the contrary, when the code generation is lower than the transmission rate, the quantization step is reduced to promote the code generation. When the buffer overflows, the quantization step is made very large and the code generation is stopped. By the above operation, a certain amount of codes are generated per time.

Next, the mask portion 22 in the embodiment will be described.

In general, many codes tend to be generated in the intraframe coding process.

Therefore, the mask portion 22 in the present embodiment.
When the fact that the intra-frame coding process has been selected is received from the process selection circuit 27 via the signal line 115, 1 in the matrix obtained from the orthogonal transformer 21 is irrespective of the quantization step g signal. Only the DC component in the 1st row is left, and the coefficients of other AC components are all masked, that is, "0". FIG. 8A shows this state.

When a signal indicating that the target block is to be encoded by either the inter-frame differential encoding process or the motion-compensated inter-frame differential encoding process is received, the quantization step g The masked area as shown in FIG. 8B is changed according to the value.

Specifically, as shown in FIG. 8B, the larger the quantization step value g, the larger the masked area, and the smaller the quantization step g value, the smaller the masked area. The reason is that the small value of the quantization step g means that the code generated up to the previous block line is small. Therefore, in this case, since it means that the orthogonal transform coefficient may be divided in multiple stages, the area to be masked is reduced. However, as shown in the figure, if the quantization step g is directed in the larger direction, the masked area is increased from the high-frequency component with less influence to the low-frequency component.

As a result, the masking is performed with the quantization step g, that is, the size corresponding to the generated code amount in the encoding process, and it is possible to suppress the excessive generation of the code in one block.

Next, the encoder 24 in the embodiment will be described.

The encoder 24 in the embodiment basically
For luminance (eg Y) data, 16x16 is 8x8
The pixel block is divided into four, and the color difference data is divided into two 8 × 8 pixels at the same position as the 16 × 16 pixel block.
6 blocks in total are generated. Two color difference blocks (for example, I and Q) are extracted by taking out a pixel block of 16 × 16 at an interval of 2 pixels both vertically and horizontally
The block size is × 8. 16x in input image
Encoding is performed with four luminance blocks and two color difference blocks as a unit of 16 blocks. The reason why the number of color difference blocks is small is that in the case of a moving image, the intervals at which each frame is displayed are instantaneous, and the influence of the color difference data in this case on humans is not so important.

The encoder 24 in the embodiment receives the signal from the selection mode signal 115 output from the process selection circuit 27 and encodes the block of interest. FIG. 9 shows a format of encoded data output based on the selection mode signal. The encoded data will be described below.

I) Interframe Differential Encoding Processing In the interframe differential encoding processing, as shown in FIG. 9A, first, an ID code indicating a processing mode (the data following the ID code is data by the interframe differential encoding processing). Code) indicating that the value of the quantization step g follows. However, according to the embodiment, the value of the quantization step g is 1
Since there is no change in the encoding of the block line, the value of the quantization step g is inserted in the encoded data of the first block of one block line. Then, the fact that the value of the quantization step g is the same is included in the IDs in the encoded data of the second and subsequent blocks, and the data of the quantization step g is not inserted. This further reduces the amount of encoded data generated. This is the reason why the value of the quantization step g is not changed in one block line. The timing of inserting the data in the quantization step g is the same in the motion compensation interframe difference encoding process and the intraframe encoding process, which will be described later.

Now, the ID code or the quantization step g
After the data, as described above, four 8 × 8 pixel blocks corresponding to the luminance data and 2 corresponding to the color difference are provided.
Out of a total of 6 blocks of 8 × 8 pixel blocks,
There is a code indicating the number of blocks that have actually been coded (that is, a significant frame difference has been recognized). This is followed by the number of encoded block data indicated by the code data. The code of each block data is composed of spectrum data and EOB (End Of Block) code. In the illustrated case, the number of significant blocks is four.

Ii) Motion Compensation Frame Differential Encoding Process In the inter-frame differential encoding process with motion compensation, as shown in FIG. 9B, first there is an ID code indicating the processing mode, and then there is the value of the quantization step g. . Next, there is a code representing a motion vector. Next, among the total of 6 blocks of 4 8 × 8 pixel blocks for luminance data and 2 8 × 8 pixel blocks for color difference data, the code is actually coded. There is a code indicating the number of blocks that have been converted.
The code for the block follows the number given here, and the code for each block consists of the spectral data and the EOB code.

Iii) Intra-frame encoding processing In the intra-frame encoding processing, as shown in FIG. 9C, first, as in the above, there is an ID code and a quantization step g, and thereafter, codes for all six blocks are arranged. . The code for each block is a code representing a fixed-length DC component (DC component), and then a code representing an AC component (AC).
Component), followed by EOB code.

In the above description, the discrete cosine transform is applied to the original image block or the difference block, and the spectrum data is encoded regardless of whether the encoding processing method for intraframe or interframe is selected. The conversion process will be performed. In each encoding process at this time, the code data shown in FIGS. 9A to 9C is generated as described above.

Therefore, for example, the luminance data 16 × 16 pixel block is almost completely flat (which means that it is almost the same), and the two color difference data 8 × 8 blocks are also flat, and the previous frame image is large. The generated code generated in each encoding process in the encoding process in the case of an image with alternating current (this situation is likely to appear when a complex image changes to a flat image due to a scene change) , As schematically shown in FIGS.

As described above, according to the first embodiment, by changing the processing mode selection method according to the size of the quantization step, it is possible to properly select the processing and to eliminate unnecessary codes. Generation and transmission can be prevented, and an efficient coding method can be provided.

Further, in the above configuration, by masking the high frequency part of the transform coefficient after the discrete cosine transform of the difference between blocks in the encoding process with a size according to the generated code amount, one block It is possible to suppress the excessive generation of codes.

[Second Embodiment] In the embodiment described above,
The generated code data format in the encoding process is basically the one schematically shown in FIGS. 10A to 10C.
Therefore, the generated code amount by the above-mentioned three coding processes, regardless of which coding process is used, generates many codes for the image with the lowest redundancy, which is to be coded from now on. There is a possibility that it will end up.

On the other hand, in addition to the encoding process only by the orthogonal transform encoding process in the above-mentioned embodiment, the encoding process in the frequency space by using the orthogonal transform and the average value data in the block in the real space are performed. Encoding processing that adaptively uses encoding, for example, an average value of 16 × 16 pixel blocks of luminance data and two corresponding color difference blocks 8 ×
By providing a new encoding process for performing the fixed length encoding process on the average value of 8 pixels, more efficient encoding process can be performed.

A second embodiment according to the present invention having the original image average value encoding processing configured as described above will be described below.

FIG. 15 shows a code format in the original image average value encoding process used in the second embodiment. The code structures of the inter-frame differential coding process, the inter-frame differential coding process with motion compensation, and the intra-frame coding process are the same as the code structures shown in FIG. 10 of the first embodiment described above. Therefore, only the code configuration in the original image average value encoding process added in the second embodiment will be described.

In the original image average value encoding process, first, a code indicating the processing mode (original image average value encoding process), and then 6
A code indicating the average value of 16 × 16 pixels of luminance block data of about a bit and a code indicating the average value of 8 × 8 pixels of two color difference blocks follow. Since these use fixed length codes, there is no EOB code used in other processing modes. That is, if the device on the decoding side detects this ID, it knows how many bits of data the EOB becomes unnecessary.

FIG. 11 is a block diagram of a moving picture coding apparatus according to the second embodiment of the present invention, which is the first block shown in FIG.
The same components as those in the embodiment are designated by the same reference numerals and detailed description thereof will be omitted.

The second embodiment is different from the first embodiment shown in FIG. 1 in that an average value calculation circuit 40 and a block data generator 41 are added, and the processing selection circuit 27 is used in the selection processing described later. In addition to the encoding process of the first embodiment, a signal indicating whether or not to perform the new average value encoding process is output to the signal line 115.

In the second embodiment, the block data input from the input signal line 103 is also sent to the average value calculation circuit 40. The average value calculation circuit 40 divides the block data input from the signal line 103 into one luminance block and two color difference blocks, calculates respective average values, and outputs the calculation result to the signal line 210. The calculation result is input to the encoder 24 and the block data generator 41.

The processing selection circuit 27 sends the processing mode via the signal line 115 when the average value coding processing described later is selected. Therefore, the encoder 24
It is possible to perform the encoding process using the average value data from the average value calculation circuit 210. Then, the block data generator 41 outputs the block data reconstructed from the average value from the average value calculation circuit 40. The output block data is stored in the previous frame memory 28 again.

Also in the second embodiment, the process selection circuit 2
The basic configuration of 7 is the same as that of FIG. 2 described above. However, the process selection control is different in this embodiment.
A determination method in the selector 1 of the second embodiment will be described below with reference to the flow chart of FIG.

In FIG. 12, first, in step S11, which is the determination A, it is checked whether or not the code generation amount is large and the overflow state is present. In the case of overflow, block data at the same position on the previous frame from the signal line 112 is selected in order to select the interframe differential encoding method.

On the other hand, if there is no overflow in step S11, the process proceeds to decision B in step S12, and the average value coding method, the intraframe coding method, or the interframe differential coding is used as the coding method. Decide whether or not to implement the method. Specifically, the signal line 107,
The determination is made based on VarOr and VarDif from 111. Based on the boundary line in the characteristic shown in FIG. 13, it is determined whether the interframe coding method, the intraframe coding method, or the average value coding method is used.

If it is determined in step S12 that intra-frame coding is to be performed, the dummy block from the signal line 108 is selected, and if it is determined to be the average value coding method, processing is performed from the signal line 115 as described above. Only the mode signal is output. On the other hand, if it is determined to be interframe coding, the process proceeds to step S13.

In step S13, the signal lines 109 and 11
Whether or not motion compensation is performed is determined by Dif and MCDif from 0. This is because the best matching block data 113 by the inter-frame differential encoding method with motion compensation is selected when there is motion compensation due to the boundary line in the characteristic shown in FIG.

On the other hand, if it is determined that there is no motion compensation,
The inter-frame differential encoding method with motion compensation is not selected, and the inter-frame encoding method is determined. In this case, the same position block on the previous frame from the signal line 112 is selected.

The selected block data is the signal line 114.
Further, the selected mode is output from the signal line 115.

When the average value encoding process is selected as the processing mode, the encoder 24 encodes the luminance and chrominance average data from the average value calculating circuit 40. May be output.

Due to the characteristics shown in FIG. 5, the signal lines 111, 1
The relationship with VarDif and VarOr from 07 is INT
If it is within the RA range, the intraframe coding method is used, and In
If it is within the ter range, the interframe differential encoding method is selected.

Further, in addition to the encoding processing only by the orthogonal transform encoding processing that has been conventionally performed, the encoding processing in the frequency space by using the orthogonal transformation and the encoding of the average value data in the block in the real space are performed. An encoding process that adaptively uses, for example, an encoding process that performs a fixed-length encoding process on an average value of a 16 × 16 pixel block of luminance data and an average value of two corresponding color difference blocks of 8 × 8 pixels. By newly providing it, more efficient encoding processing can be performed.

The output coded data according to the second embodiment has four types shown in FIG. 15 in addition to FIGS. 10A to 10C.

In the above configuration, by changing the processing mode selection method according to the size of the quantization step, it is possible to appropriately select the processing and generate unnecessary codes.
It is possible to prevent transmission and provide an efficient encoding method.

The present invention may be applied to a system composed of a plurality of devices or an apparatus composed of a single device. Further, it goes without saying that the present invention can also be applied to the case where it is achieved by supplying a program to a system or an apparatus.

FIG. 21 is a block diagram showing the overall apparatus configuration including the encoder of each embodiment of the present invention.

In FIG. 21, reference numeral 1000 denotes an input device such as a video camera for sequentially inputting image information for each screen.
001 is a frame memory that stores image information from the input device for each screen, 1002 is a block cutout circuit that cuts out the image information stored in the frame memory 1001 for each 8 × 8 pixel block, and 1003 is data from the block cutout circuit 1002. An encoding unit that encodes the block 103 into variable-length encoded data, for example, FIG. 1 and FIG.
Is shown in. 1004 is an encoder, (i)
If the quantization step g is changed, one of the values and g, (ii) INTRA / MC / inter determination result,
(iii) In the case of the inter-frame differential coding method with motion compensation, vector data representing the relative one between the block of the best matching and the block to be coded, (iv) other code parameters are coded. A multiplexer 1005 combines the variable-length code data from the encoder 23 and the code data from the encoder 1004 to sequentially control the control circuit 10.
Output to 06.

The control circuit 1006, as shown in FIG.
The input code data from the multiplexer 1005 is counted, and the code buffer 1007, which will be described later, transmits the code data.
The coding in the coding unit 1003 is controlled so that the code buffer 1007 does not overflow depending on the amount of coded data output according to the capability of 009.

This control is performed by the quantization step g101 and the overflow signal 102. The former becomes larger as the amount of code data stored in the code buffer 1007 increases as described above, and the latter becomes stored in the code buffer 1007. This indicates when the code data amount exceeds a predetermined value.

A code buffer 1007 temporarily stores code data until it is transmitted from the transmission circuit 1008. Reference numeral 1009 denotes a transmission path, which has a capability of transmitting a certain fixed amount of code data per unit time.

The above-mentioned input device is not limited to a video camera, but may be a device for inputting moving image data created by software of a computer.

Further, the coding is not limited to the above DCT, and the coding may be performed by using an orthogonal transform such as Hadamard transform.

The present invention is not limited to the above-mentioned embodiments, but can be variously modified and applied within the scope of the claims.

[0128]

As described above, according to the present invention, it is unnecessary to change the criterion for selecting the coding process based on the transition of the coded data amount that is occurring for at least one frame. It is possible to prevent the generation of such a code.

Further, according to another invention, the masking size after the orthogonal transform is changed based on the transition of the coded data amount which is occurring for at least one frame, so that in one block. It is possible to suppress excessive generation of codes.

Further, according to another invention, by adding a new encoding process in addition to the ordinary encoding process, the encoding process can be performed more efficiently.

[0131]

[Brief description of drawings]

FIG. 1 is a block diagram of a moving image coding apparatus according to an embodiment.

FIG. 2 is a detailed block diagram of a process selection circuit in FIG.

FIG. 3 is a flowchart showing selection processing contents of a processing selection circuit.

[Fig. 4] Fig. 4 is a diagram illustrating a criterion for determining whether or not to perform an interframe differential encoding method with motion compensation.

FIG. 5 is a diagram showing a criterion for deciding whether to perform an intraframe coding method or select an interframe differential coding method.

FIG. 6 is a diagram showing another criterion for deciding whether or not to perform the inter-frame differential encoding method with motion compensation according to the present embodiment.

FIG. 7 shows whether the intraframe coding method according to the present embodiment is performed,
FIG. 11 is a diagram showing another criterion for determining whether to select the interframe differential encoding method.

FIG. 8A is a diagram showing the mask processing contents of the mask unit in the present embodiment.

FIG. 8B is a diagram showing the mask processing contents of the mask unit in the present embodiment.

FIG. 9A is a diagram schematically showing a generated code in a certain state in the encoding process in the example.

FIG. 9B is a diagram schematically showing a generated code in a certain state in the encoding process in the example.

FIG. 9C is a diagram schematically showing a generated code in a certain state in the encoding process in the example.

FIG. 10A is a diagram schematically showing the generated code amount in the encoding process in the example.

FIG. 10B is a diagram schematically showing the generated code amount in the encoding process in the embodiment.

FIG. 10C is a diagram schematically showing a generated code amount in the encoding process in the example.

FIG. 11 is a block diagram of a moving picture coding apparatus in the second embodiment.

FIG. 12 is a flowchart showing a process selection method of an encoding method according to the second embodiment.

FIG. 13 is a diagram showing a criterion for determining whether or not to perform the average value coding method and the intraframe coding method according to the second embodiment.

FIG. 14 is a diagram showing a criterion for selectively determining whether or not motion compensation is to be performed in the interframe differential encoding method according to the second embodiment.

FIG. 15 is a diagram showing a generated code amount in an original image average value encoding process used in the second embodiment.

FIG. 16 is a flowchart showing an example of a process selection method of an encoding method.

[Fig. 17] Fig. 17 is a diagram for describing a criterion for determining whether to perform the intraframe coding method or select the interframe differential coding method with motion compensation.

[Fig. 18] Fig. 18 is a diagram illustrating a criterion for determining whether to perform an intraframe coding method or select an interframe differential coding method.

FIG. 19 is a block diagram showing a moving image encoding device.

FIG. 20 is a diagram showing a specific example of a circuit for generating a quantization step g signal in the embodiment.

FIG. 21 is a block diagram of the entire apparatus including the moving image encoder in the first and second embodiments.

[Explanation of symbols]

 1 selector 2 interframe difference calculation and motion compensation circuit 3 intraframe correlation calculator 21, 31 discrete cosine transform circuit 22, 32 transform coefficient mask circuit 23, 33 quantizer 24, 34 encoder 25, 35 dequantizer 26, 36 Inverse converter of discrete cosine transform 27 Process selection circuit 28, 38 Previous frame Frame memory 37 Mode determiner 40 Average value calculation circuit 41 Block data generator

Claims (18)

[Claims] 1. A coding method for coding individual pixel blocks in units of pixel blocks of a predetermined size that form a frame image, the correlation between a pixel block in a frame of interest and a corresponding block in a previous frame. Based on the relationship, a determination process for determining whether the intra-frame coding process or the inter-frame differential coding process is performed on the pixel block of interest, and based on the previously encoded data amount, the determination process in the determination process is performed. And an adjusting step for adjusting a judgment criterion. 2. The inter-frame differential encoding processing includes inter-frame differential encoding processing with motion compensation and inter-frame differential encoding processing without motion compensation, wherein the determination step is to encode. The process of calculating the difference value of each of the intra-frame coding process, the inter-frame difference coding process with motion compensation, and the inter-frame difference coding process without motion compensation for the block of interest, and the process of calculating the difference value The process of determining the coding process of the block of interest depending on which of the first boundary lines in the coordinate space the respective values can take, and the position selected by the inter-frame difference coding process with motion compensation. Variance value between the block in the frame and the block of interest,
And a step of determining the encoding process of the target block depending on which of the second boundary lines in the respective possible coordinate spaces the position defined by the variance value of the target pixel block itself is, 2. The encoding method according to claim 1, wherein the first and second boundary lines are moved and adjusted. 3. A coding device for coding individual pixel blocks in units of pixel blocks of a predetermined size that form a frame image, the correlation between a pixel block in a frame of interest and a corresponding block in a previous frame. Based on the relationship, the determination means for determining whether the intra-frame coding process or the inter-frame differential coding process is performed on the pixel block of interest, and based on the coded data amount in the past, An encoding device, comprising: an adjusting unit that adjusts a determination criterion. 4. The inter-frame differential encoding process includes an inter-frame differential encoding process with motion compensation and an inter-frame differential encoding process without motion compensation, and the determination means is about to encode. A means for calculating the difference value of each of the intra-frame coding process, the inter-frame difference coding process with motion compensation, and the inter-frame difference coding process without motion compensation for the block of interest, and the specified difference value. The first judgment means for judging the coding process of the target block depending on which of the first boundary lines in the coordinate space the respective values can take, and the inter-frame differential coding process with motion compensation. Variance value between the selected block in the previous frame and the block of interest,
And a second determination unit that determines the encoding process of the target block depending on which of the second boundary lines in the coordinate space each of which the position defined by the variance value of the target pixel block itself is. The encoding device according to claim 3, wherein the adjusting unit moves and adjusts the first and second boundary lines. 5. A pixel block of a predetermined size forming a frame image is selected from one of encoding processes selected based on a correlation with a corresponding block of a previous frame,
In the encoding method for encoding, the orthogonal transformation process of orthogonally transforming the block of interest to which the difference information based on the selected encoding process is added, and the block data that has been orthogonally transformed based on the amount of encoded data up to now And an adjusting step for adjusting a masking ratio of the alternating current component therein. 6. The adjustment step is characterized in that the larger the encoded data amount up to now is, the larger the masking rate of the AC component of the target block data of the orthogonal transform is. The encoding method according to item. 7. The adjustment step is masked in a region that expands from a high frequency component to a low frequency component in the AC component as the encoded data amount up to the previous time increases. The encoding method according to the sixth item. 8. A pixel block of a predetermined size forming a frame image is selected from one of encoding processes selected based on a correlation with a corresponding block of a previous frame,
In an encoding device for encoding, orthogonal conversion means for orthogonally converting a block of interest to which difference information based on the selected encoding processing is added, and the block data that has been orthogonally converted based on the amount of coded data up to now. An encoding device, comprising: an adjusting unit that adjusts a masking ratio of an alternating current component therein. 9. The adjusting means increases the masking rate of the AC component of the block data of interest in the orthogonal transform as the encoded data amount up to now increases. The encoding device according to the item. 10. The adjusting means masks in a region that expands from a high-frequency component to a low-frequency component in the alternating-current component as the amount of encoded data up to now increases. The encoding device according to the ninth item. 11. A coding method for coding a pixel block of a predetermined size forming a frame image by selecting one of coding processes selected based on a correlation with a corresponding block of a previous frame. , The first conversion step of converting the pixel block to which the difference information based on the selected encoding process is added in the frequency space, and averaging in the type information unit included in the raw pixel block, A second conversion step of converting to average value data, a selection step of adaptively selecting the first conversion step and the second conversion step, and encoding the selected converted pixel block An encoding method characterized by the above. 12. The encoding process corresponding to the first conversion process includes inter-frame differential encoding process and intra-frame encoding process, and the selection process includes encoding by the second conversion process. The encoding method according to claim 11, wherein one is selected from among processing, the interframe differential encoding processing, and the intraframe encoding processing. 13. The selecting step includes a step of calculating a variance value of a block of interest to be encoded and a variance value of a block in a previous frame selected by interframe difference encoding processing of the block of interest. Of the second conversion process, the interframe difference encoding process, and the intraframe encoding process based on the calculated dispersion value corresponding to the interframe differential encoding process and the calculated dispersion value corresponding to the pixel block of interest. The encoding method according to claim 12, further comprising a step of selecting one of them. 14. The inter-frame differential encoding processing comprises:
An inter-frame differential encoding process with motion compensation and an inter-frame differential encoding process without motion compensation are included. The selection process is based on the two differences when the inter-frame differential encoding process is selected. 14. The encoding method according to claim 13, wherein one of the two is selected. 15. A coding device for coding a pixel block of a predetermined size which constitutes a frame image by selecting one of coding processes selected based on a correlation with a corresponding block of a previous frame. , A first conversion means for converting the pixel block to which the difference information based on the selected encoding process is added in the frequency space, and averaging in the type information unit included in the raw pixel block to obtain the entire pixel block. A second conversion unit that converts the average value data, and a selection unit that adaptively selects the first conversion unit and the second conversion unit are provided, and the selected pixel block after conversion is encoded. An encoding device characterized by the above. 16. The encoding processing corresponding to the first converting means includes inter-frame differential encoding processing and intra-frame encoding processing, and the selecting means corresponds to the second converting means. The encoding device according to claim 15, wherein one of the encoding process, the inter-frame difference encoding process, and the intra-frame encoding process is selected. 17. The selecting means calculates a variance value of a block of interest to be encoded and a variance value of a block in a previous frame selected by interframe difference encoding processing of the block of interest. Of the second conversion process, the interframe difference encoding process, and the intraframe encoding process based on the calculated dispersion value corresponding to the interframe differential encoding process and the calculated dispersion value corresponding to the pixel block of interest. The encoding device according to claim 16, further comprising means for selecting one of them. 18. The inter-frame differential encoding processing comprises:
An inter-frame differential encoding process with motion compensation and an inter-frame differential encoding process without motion compensation are included. When the inter-frame differential encoding process is selected, the selecting means further determines based on the two differences. 18. The encoding apparatus according to claim 17, wherein either one of them is selected.