JP3348776B2 - Moving picture coding apparatus and method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

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

[0002]

2. Description of the Related Art ISDN Providing Digital Data Transmission
With the spread of lines, the use of TV conference terminals and remote image monitoring using a moving image encoding device that transmits moving images at a low bit rate is rapidly expanding. As a compression method of a moving picture encoding apparatus for transmitting moving pictures, for example, ITU-T Recommendation H.264 261 is used. In this method, image compression is performed to reduce redundancy in the time axis direction by inter-frame coding and to reduce spatial redundancy by discrete cosine transform coding, quantization, and variable-length coding.

[0003]

In this method, when the subject moves, the amount of inter-frame coding increases. Therefore, in order to transmit the data through an ISDN line having a fixed bit rate, the quantization is made coarse to reduce the amount of generated information. It is necessary to hold down. When quantization is made coarse, quantization noise increases, resulting in a deteriorated image such as block distortion or mosquito noise. In particular, after the subject moves, these deteriorations continue for several frames, resulting in a dirty image.

FIG. 4 shows an example of a conventional moving picture coding apparatus, in which an input moving picture signal is divided into blocks by a block or a circuit 909, and an input moving picture divided into small blocks and local decoding one frame before. An image difference is taken by a subtractor 901, the inter-frame difference signal is subjected to discrete cosine transform by a DCT circuit 902, the transform coefficient is quantized by a quantizer 903, and the quantized transform coefficient is encoded by an encoder 908. And compresses the moving image.

Here, it is assumed that there is an input moving image as shown in FIG. The “squares” in FIG. 3 are moving object images, “triangles” and “
A circle 'is a stationary background image, and an example in which one screen is divided into 16 blocks will be described. When the 'square' of the frame (a) moves to the position of 'triangle' in the next frame (b), the difference between the frames is ('square'-'triangle'), and the inter-frame difference signal is a DCT. conversion·
The inverse quantizer 90 simultaneously performs the quantization and transmission.
4. The inverse DCT circuit 905 performs inverse quantization and inverse discrete cosine transform, and stores the result in the frame memory 907 as a locally decoded image. In the next frame (c),
The 'rectangular' moving object moves and a 'triangle' hidden by the 'quadrangle' appears. In the conventional example, a ('triangle'-'square') inter-frame difference signal is encoded. Therefore,
Again, ('triangle'-'square') must be encoded, and the degradation continues at this part for several frames, resulting in a dirty image.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an image coding apparatus of an inter-frame coding system in which a stain on an image after a subject has moved is quickly eliminated.

[0007]

According to the present invention, there is provided a moving picture coding apparatus comprising: a subtracter for obtaining a difference between an input image signal and an output of a selection circuit to output a difference image signal; A spatial redundancy reducing unit for reducing redundancy, an inverse transforming unit for inversely transforming an output of the spatial redundancy reducing unit, an output of the inverse transforming unit and an output of the selection circuit are added to obtain a current local An adder for outputting a decoded image signal; delay means for delaying the current local decoded image signal by a predetermined number of frames to output a reference image signal; storage means for storing a background image signal; A selection circuit that selects a signal or an output signal of the storage unit; and a signal selected by the selection circuit based on the input image signal, the reference image signal, and the background image signal output by the storage unit. (1) determining means, based on the input image signal, the reference image signal, and the background image signal output by the storage means, wherein the storage means loads the current local decoded image signal into the storage means as the background image signal Second determining means for determining whether or not this is the case.

[0008] Further, a moving picture coding apparatus according to the present invention comprises:
In the above moving picture encoding apparatus, the first determination unit may include a value calculated based on a difference between the input image signal and the reference image signal, and a value output by the input image signal and the storage unit. A signal to be selected by the selection circuit is determined by comparing a value calculated based on a difference with a background image signal.

Further, the moving picture coding apparatus according to the present invention
In the above moving picture encoding apparatus, the second determination means may determine whether the storage means fetches a current locally decoded image signal as the background image signal based on a determination result of the first determination means over a plurality of frames. It is characterized in that it is determined whether or not it is.

[0010] Further, the moving picture coding apparatus according to the present invention comprises:
Background image signal storage means for storing a background image signal; means for determining whether a current input image signal is a background image signal; and when it is determined that the current input signal is the background image signal. Means for inter-frame encoding using a past background image signal stored in the background image signal storage means in place of the reference image signal, and the current input signal is determined to be the background image signal And a means for storing the local decoded image signal at that time in the background image signal storage means.

A moving picture coding method according to the present invention includes a subtraction step of obtaining a difference between an input image signal and an output in a selection step to output a difference image signal, and reducing the spatial redundancy of the difference image signal. A spatial redundancy reduction step, an inverse transformation step for inversely transforming the output of the spatial redundancy reduction step, and adding the output of the inverse transformation step and the output of the selection step to output a current local decoded image signal An adding step of delaying the current locally decoded image signal by a predetermined number of frames and outputting a reference image signal; and a writing step of taking the current locally decoded image signal as a background image signal into storage means. Selecting the reference decoded image signal or the output signal of the storage step, the input image signal, the reference image signal and the The determined signal for selecting in said selecting step based on the background image signal 憶 means outputs 1
And a second determination step of determining whether to execute the writing step based on the input image signal, the reference image signal, and the background image signal output by the storage step. Features.

Further, the moving picture coding method according to the present invention comprises:
In the above moving image encoding method, in the first determining step, a value calculated based on a difference between the input image signal and the reference image signal, and a value output from the input image signal and the storage unit are output. The signal to be selected in the selection step is determined by comparing a value calculated based on a difference with a background image signal.

Further, the moving picture coding method according to the present invention comprises:
In the above moving image encoding method, in the second determining step, it is determined whether or not to execute the writing step in the storage unit based on a determination result over a plurality of frames in the first determining step. Features.

A moving image encoding method according to the present invention comprises the steps of: determining whether a current input image signal is a background image signal; and determining whether the current input signal is the background image signal. Inter-frame encoding using a past background image signal stored in a background image signal storage means instead of a reference image signal, and when it is determined that the current input signal is the background image signal And storing the locally decoded image signal at that time in the background image signal storage means.

A computer-readable recording medium according to the present invention is characterized in that a program for causing the computer to function as the above-described image encoding apparatus is recorded.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention relates to a moving picture coding apparatus which divides an input moving picture signal into small blocks and compresses the image in units of the divided small blocks by an inter-frame discrete cosine transform coding method. A frame memory for local decoding and a background memory for storing a background image are prepared. If an image of a region where a moving object becomes a background image as a result of moving a certain object, the moving object is set for each block. Judge whether the block is a background block or not, and if a moving object block changes to a background block, find the difference between frames between the input moving image and the image in the background memory to quickly converge the dirt after the subject moves. A moving picture coding apparatus is provided.

Referring to the example of FIG. 3, in the conventional example, an inter-frame difference signal of (`triangle`-`square`) is encoded in frame (c). By taking the inter-frame difference between the 'triangle' image stored in the background memory and the input image instead of the 'square' local decoded image, the ('triangle'-'triangle') inter-frame difference signal is encoded. Will be done. Therefore, it is possible to greatly reduce the difference signal.

FIG. 1 shows an embodiment of a moving picture coding apparatus.

Referring to FIG. 1, the configuration of this embodiment is as follows.
The input moving image signal 99 is converted into, for example, 16
The block is divided into × 16 small blocks, and the local decoded image signal 108 or the background image signal 109 one frame before is divided by the selection circuit 18.
Signal 106 and input video signal 100 selected from
Is obtained by a subtractor 10, and a discrete cosine transform by a DCT circuit 11, a quantization 12 by a quantizer 12, and an encoding by an encoder 20 are performed on the inter-frame difference signal 101 to obtain an encoded image signal. 112 is obtained and output.

At the same time, an image signal 105 that has been subjected to inverse quantization 13 by an inverse quantizer 13 and inverse discrete cosine transform by an inverse DCT circuit 14 to the quantized transform coefficient 103 and an image signal 106 from a selection circuit 18 are added. Thus, a locally decoded image signal 107 is obtained. This locally decoded image signal 107 is stored in the frame memory 16 for encoding the next frame and at the same time, in accordance with a background memory update instruction signal 111 from a background block determination circuit 19 described later.
Is stored in the background memory 17.

The background block judging circuit 19 outputs the local decoded image signal 108 from the frame memory 16 one frame before.
And the background image signal 109 stored in the background memory 17 and the input moving image signal 100, and outputs an instruction signal 110 for selecting an image to be input to the subtractor 10 to the selection circuit 18 and simultaneously outputs the background block flag 110. Circuit.

Next, the operation of the embodiment of the present invention will be described in detail.

The input moving image signal 100 is, for example, 16 × 1
The image signal is divided into six small blocks, and the difference between the divided small blocks and the image signal 106 from the selection circuit 18 is calculated by the subtracter 10 to be converted into an inter-frame signal. That is, the image signal 101 with reduced redundancy in the time domain can be obtained. The image signal 101 from which the redundancy in the time domain has been removed is represented by D
The spatial redundancy is removed by the processing of the CT circuit 11 and the quantizer 12. The quantized transform code 103 output from the quantizer 12 is subjected to variable-length encoding by the encoder 20 and output as a variable-length output code 112. After the quantized transform code 103 is subjected to inverse quantization by the inverse quantizer 13 and inverse discrete cosine transform by the inverse DCT circuit 13, the quantized transform code 103 is selected from the local decoded image 108 or the background image signal 109 of the previous frame. The signal is added to the image signal 106 selected by the circuit 18, and as a result, the local decoded image 107
Is obtained. The local decoded image signal 107 is stored in the frame memory 16 for encoding the next frame, and is also stored in the background memory 17 according to the background memory update instruction signal 111 from the background block determination circuit 19. The background block determination circuit 19 receives the local decoded image signal 108 one frame before from the frame memory 16, the background image signal 109 stored in the background memory and the input moving image signal 100, and outputs the image to the subtracter 10. Is output to the selection circuit 18 and the background block flag 110 is output at the same time.

Next, details of the background block determination circuit 19 will be described with reference to FIG.

An input video signal 100 divided into 16 × 16 small blocks and a frame memory output video signal 108
Is input, the absolute difference value of each pixel is obtained by the subtractor 20 and the absolute value circuit 22, and the sum of the absolute difference values of the pixels in the block is obtained by the adder 24 to obtain the difference power 204 (P
Find d). Similarly, the difference power 205 (Pb) between the input moving image signal 100 and the background memory output image signal 109 is obtained by the subtractor 21, the absolute value circuit 23 and the adder 25. The respective difference powers Pd and Pb are compared with the comparator 26.
And outputs a background block flag 110 according to the following criteria.

(1) When Pd ≦ Pb The background block flag = 0 is output, and an instruction to select the local decoded image signal 108 from the frame memory 16 is given to the selection circuit 18.

(2) In the case of Pd> Pb The background block flag = 1 is output, and the selection circuit 18 is instructed to select the background image signal 109 from the background memory 17.

The absolute value circuits 22 and 23 may be replaced by a square circuit.

Next, the background memory update determination circuit 28 sends a background memory update instruction signal 111 to the background memory 17 according to the following criteria based on the determination result 110 from the comparator 26 and the determination result 206 one frame before. Output.

(1) In the case where the current decision result 110 = 1 and the decision result 206 immediately before the previous frame = 1 = 1 From the result that the difference power from the locally decoded image is small for two consecutive frames, the corresponding small block is a background block. And outputs the background memory update instruction 111 = 1 to instruct the background memory to fetch the local decoded image signal 107.

(2) Cases other than the condition of (1) It is determined that the image stored in the background memory is valid as the background, a background memory update instruction 111 = 0 is output, and the local decoded image 107 is fetched into the background memory. Instruct not to.

In the above example, the case where two frames are continuous is assumed. However, by providing a plurality of one-frame delay units 27, the condition of continuation of a plurality of frames can be realized, and the accuracy of determining whether or not a background is present can be improved. .

In the above embodiment, the inter-frame difference is subjected to the discrete cosine transform by the DCT circuit 11 and quantized by the quantizer 12 to reduce the redundancy in the spatial direction, and the corresponding inverse quantization is performed. A circuit 14 and an inverse quantizer 13 are also provided. Further, even if there is no space-direction redundancy reduction means for reducing the redundancy in the space direction and a means corresponding thereto, the effects of the present invention are not impaired.

In the above embodiment, the inter-frame prediction is performed based on the image signal of the previous frame.
The present invention is not limited to this, and the present invention can also be applied to a method of performing inter-frame prediction based on an image signal one field before or based on a signal several frames before. Further, the present invention can be applied to an inter-frame coding system that performs inter-frame prediction in two directions, such as the MPEG system.

Further, the apparatus according to the above-described embodiment uses a computer for each means (blocking circuit 21, subtractor 10, DCT circuit 11, quantizer 12, inverse quantizer 13, inverse DCT circuit 14, Adder 15, selection circuit 18,
A computer that reads and executes the program from a recording medium such as a CD-ROM in which a program for functioning as the background block determination circuit 19, the encoder 20, the frame memory 16, and the background memory 17) is recorded. Can also.

At this time, usually, the CPU of the computer is used.
After the program is once transferred from the recording medium to the main memory by the operating system, the program is read from the main memory and executed.

The above program may be generated by a text file in which an interpreter-format instruction is described, and an interpreter which reads this file and converts it into an execution instruction.

[0038]

As described above, according to the present invention,
A background memory and a background block determination circuit for storing a background image signal are provided, and when it is determined that the input image signal is the background image signal, the input signal and the background signal are used instead of the local decoded signal. , It is possible to reduce the information generated in the inter-frame coding, and in particular, quickly converge the dirt after the subject has moved, thereby enabling clear moving image transmission.

[Brief description of the drawings]

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

FIG. 2 is a block diagram illustrating a configuration example of a background block determination circuit in FIG. 1;

FIG. 3 is a diagram illustrating an example of an input image of the image encoding device.

FIG. 4 is a block diagram illustrating a configuration of a conventional image encoding device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Subtractor 11 DCT circuit 12 Quantizer 13 Inverse quantizer 14 Inverse DCT circuit 15 Adder 18 Selection circuit 19 Background block determination circuit 20 Encoder 21 Blocking circuit

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N ^7/ 24-7/68

Claims (3)

(57) [Claims] 1. A subtractor for taking a difference between an input image signal and an output of a selection circuit and outputting a difference image signal , a DCT unit for performing a discrete cosine transform of the difference image signal , and an inverse discrete cosine output of the DCT unit Inverse DC to convert
T means; an adder for adding the output of the inverse DCT means and the output of the selection circuit to output a current locally decoded image signal; and a reference image which delays the current locally decoded image signal by a predetermined number of frames. delay means for outputting a signal, a storage means for storing signals determined to be the background image, the output signal of said reference decoded image signal and the storage means
A selection circuit for performing one of the selection of Chi, based on a difference between the reference image signal and the input image signal
The input image signal and the storage means
A signal that was determined to be the background image in the past to be output;
By comparing with the value calculated based on the difference of
In the selection, the reference decoded image signal and the memory
Decide which of the stage output signals to select
Decision signal generating means for generating a constant signal , wherein the decision signal is the current selection circuit
Indicates that the output of the storage means is to be selected, and
The selection circuit is a constant signal and the past one is the memory signal.
When the output of the stage is selected,
Means for converting the current locally decoded image signal with a new said background image
Determined to be stored as a signal determined to be present
Determination means for moving picture coding apparatus comprising: a city. Wherein the input image signal and a subtraction step of outputting a differential image signal takes the difference between the output of the selection step, the DCT step of discrete cosine transforming said differential image signal, inverse discrete output of the DCT step Inverse cosine transform
A DCT step, an adding step of adding an output of the inverse DCT step and an output of the selecting step to output a current locally decoded image signal, and converting the current locally decoded image signal into a predetermined number of frames.
Writing a first write step <br/> flop writing to delay means Ru delaying a signal, the signal is determined to be a background image in the storage unit
A second writing step, before read out from the reference decoded image signal and the storage means is a local decoded image signal delayed by said delay means
Serial read signal is signal determined to be a background image
A selection step of performing one of the selection of Chi, based on a difference between the reference image signal and the input image signal
The input image signal and the readout signal.
By comparing with the value calculated based on the difference of
In the selecting step, the reference decoded image signal and the
Deciding which of the read signals to select
And the result of the determining step , wherein the current one is the selected one.
Indicates that the read signal is selected in the step.
One of the results of the decision step and the past
Indicates that the read signal is selected in the selection step.
When the current local decoded image signal is stored in the storage means.
Write as a signal determined to be the background image
A determining step of determining that the video should be taken. 3. A computer-readable recording medium on which a program for causing a computer to execute the image encoding method according to claim 2 is recorded.