JPH09130808A - Method and device for compressing dynamic image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method and device for compressing dynamic image efficiency search-processing a motion vector by judging an optimum motion vector in the middle step of search and simplifying search-processing so as to average a search processing amount corresponding to the difference of picture types in the method and device for compressing dynamic image searching plural kinds of motion vectors by multi-step search by dividing hierarchy from rough search to fine search. SOLUTION: A judging step switching unit 30 switches a judging step so as to judge a motion vector based on the motion vector at a final step searched by first and second motion vector searching units 32 and 34 in the case of an image P and to judge the motion vector based on the motion vector at a first step searched by the first motion vector searching unit 32 in the case of an image B.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture compression method and apparatus for compressing a digital moving picture, and more particularly to a moving picture in which each image forming the digital moving picture is coded based on a plurality of types of prediction images. A compression method and apparatus.

[0002]

2. Description of the Related Art In recent years, multimedia which expresses different information such as characters, figures, voices, and images by digital data, and integrates these media and handles them in a unified manner has attracted attention in recent years. As an audio / video encoding system compatible with this multimedia, ISO / IEC MPEG (Mo
(ving Picture Experts Group) 1 and MPEG2 have each been recognized as international standards.

In these encoding methods, first, for each screen (hereinafter, also referred to as a picture) forming a moving image, luminance information and color difference information contained in pixels forming the screen are based on a predetermined sampling frequency. Are extracted in the horizontal scanning direction and the vertical scanning direction, so-called 4: 2: 0, 4: 2: 2.
Alternatively, the image format is converted to 4: 4: 4. Next, each converted screen is divided into, for example, rectangular blocks of 16 pixels × 16 pixels (hereinafter referred to as macroblocks). Then, for each of these macroblocks, a similar block similar in size to the macroblock of the current image is searched from the search window of the reference image, and a motion vector indicating the spatial distance and orientation to the searched similar block and Difference information from the searched similar block is calculated and motion compensation is performed. Next, these pieces of difference information are converted to DCT (Discrete Cosine Transform);
Discrete Cosine Transform), quantization and variable length coding. When the difference information is compressed in this way, it can be compressed more efficiently than when the original image itself is compressed.

As shown in FIG. 4, in MPEG, three picture types, I picture, P picture and B picture, are defined. I picture (Intra-Picture
Intra-frame predictive coded image is an intra (In) image in which all macroblocks in the picture do not refer to
tra; intra-frame coding) block.
I-pictures are periodically provided to compress and reproduce images of other picture types, and are also used for random screen access. P picture (Predictive-Pic
ture; forward predictive coded image) is a forward I in time.
Is coded with reference to a picture or P picture,
May include intra blocks. B picture (Bidi
(rectionally predictive-Picture; bidirectional predictive coded image) is coded with reference to an I picture or P picture in the forward direction and backward direction in time, and may include an intra block. However, the B picture is not used as another reference image. Further, the I picture and the P picture are temporally encoded in the same order as the original image, but the B picture is inserted between the I picture and the P picture after being encoded.

For example, in FIG. 4A, the period M = 2, the I picture is generated first, and then the P picture is generated every second picture and the generated P picture is generated.
One B picture is generated from the picture or I picture and P picture. FIG.
In (b), the period M = 3, an I picture is first generated, a P picture is generated every third picture, and two B pictures are generated from the generated P picture or I picture and P picture. Is being generated.

Further, while MPEG1 handles only progressive scan type images, MPEG2 handles encoding of interlaced scan type images. The interlaced scanning method is different from the sequential scanning method in which vertical scanning is simply performed line by line, and vertical scanning is performed by skipping every predetermined number of scanning lines, and a frame is generated according to the number of scanning corresponding to the number of skipped scanning lines. It is what constitutes. For example, in the 2: 1 interlaced scanning method applied to television images and the like, one frame is composed of two fields, a field composed of odd scanning lines and a field composed of even scanning lines. Is performed, and then the other field is scanned. This interlaced scanning system reduces the signal bandwidth, increases the number of scans of the entire screen, and reduces image flickering without substantially reducing the number of scanning lines.

An image of the interlaced scanning system is provided with both a frame structure in which a frame is a unit of coding and a field structure in which a field is a unit of coding.
Further, the prediction methods include a frame prediction method and a field prediction method. However, in the case of the field structure, the frame prediction method cannot be used. First, an example of the prediction method in the frame structure will be described.

As shown in FIG. 5, the current image frame 110
The first field 111 of the current image consisting of odd scan lines
And the second field 1 of the current image consisting of even scan lines
It is assumed that the reference image frame 120 is composed of 12 and the reference image frame 120 is composed of a first field 121 composed of odd scanning lines and a second field 122 composed of even scanning lines, and the current image frame 110 is predicted from the reference image frame 120. Also, it is assumed that the person image A shown by diagonal lines is moving from the lower left to the upper right of the screen.

The frame prediction method in the frame structure is to predict the current image frame 110 from the reference image frame 120 by the motion vector MVA. In the field prediction method in the frame structure, the current image first field 111 is moved from the reference image first field 121 or the reference image second field 122 to the motion vector M.
In addition to the prediction by VA1, the current image second field 112 is predicted by the motion vector MVA2 from the reference image first field 121 or the reference image second field 122, and the frame 110 is predicted by combining the two predicted fields. .

Here, as shown in FIG. 6, the current image is time n, the reference image is time (n-1), each pixel of the odd scan lines of the current image and the reference image is represented by a white circle, and the current image and Each pixel of the even scan line of the reference image is represented by a black circle,
Assuming that the current image frame block 210 of 8 pixels vertically is composed of the first field block 211 of current image of 4 pixels vertically and the second field block 212 of current image of 4 pixels vertically, the motion vector MVA is
10 pixel data and a plurality of frame candidate blocks 2
It is calculated based on 20 pixel data.

Further, the motion vector MVA1 is the current image first
The motion vector MVA2 is obtained based on the pixel data of the field block 211 and the pixel data of the plurality of first field candidate blocks 221. The motion vector MVA2 is the pixel data of the current image second field block 212 and the pixel data of the second field candidate blocks 222. Is calculated based on.

In reality, each of the frame block 210 and the frame candidate block 220 is 16 pixels × 16.
Each of the first field block 211, the second field block 212, the first field candidate field 221, and the second field candidate block 222 is a macro block of pixels, and each is composed of a block of 16 pixels × 8 pixels.

As described above, in the frame structure, one motion vector MVA by the frame prediction method and two motion vectors MVA1, MVA2 by the field prediction method are obtained, and the optimum motion vector is selected from these motion vectors. To be selected. Next, an example of the field prediction method in the field structure will be described. The field prediction method in the field structure further includes a 16 × 16 field prediction method and a 16 × 8 field prediction method.

As shown in FIG. 7, for example, the current image field 310 is predicted from the reference image field 320. Further, it is assumed that the person image B shown by diagonal lines is moving from the lower left to the upper right of the screen. Here, as shown in FIG. 8, the current image field 310 is time n, the reference image field 320 is time (n−1), each pixel forming the current image field 310 is represented by a white circle, and the reference image field 320 is represented by Each of the constituent pixels is represented by a black circle. In addition, a current image field block 410 of four vertical pixels in the current image field 310 constitutes a first half of the current image field block 410 and a second half of the current image field block 410. It is assumed to be composed of a segment block 412.

In the 16 × 16 field prediction method, the current image field 310 is changed from the reference image field 320 by the field motion vector MVB obtained based on the pixel data of the current image field block 410 and the pixel data of the plurality of field candidate blocks 420. To predict. On the other hand, in the 16 × 8 field prediction method, the first segment motion vector MVB1 and the current image second segment obtained based on the pixel data of the current image first segment block 411 and the pixel data of the plurality of first segment candidate blocks 421 are used. Block 412
Pixel data and a plurality of second segment candidate blocks 42
The reference image field 320 according to the second segment motion vector MVB2 obtained based on the pixel data of 2
From the current image field 310.

In practice, the field block 410 and the field candidate block 420 are each 16 pixels ×
It is a macroblock of 16 pixels, and each of the first segment block 411, the second segment block 412, the first segment candidate field 421 and the second segment candidate block 422 is composed of a block of 16 pixels × 8 pixels.

As described above, in the field structure, 16 × 1
One field motion vector MVB by the 6-field prediction method and two first and second segment motion vectors MVB1, MV by the 16 × 8 field prediction method
B2 is obtained, and the optimum motion vector is selected from these motion vectors. As described above, in the case of the frame structure, the optimum motion vector is selected from one frame motion vector and two field motion vectors, and in the case of the field structure, one field motion vector and two field motion vectors are selected. By selecting the optimum motion vector from the segment motion vectors, the compression rate of the information amount is further improved.

By the way, in general, a huge amount of calculation and memory access are required to search for a motion vector, so the processing capacity for searching for a motion vector is a major factor in determining the processing capacity of the entire coding apparatus. As a method of improving the processing capability of searching for this motion vector, for example, NTC81 (National Telecommunication Confere
nce 81) 1981, G5.3.1-G5.3.5, "Moti.
on compensated interframe coding for video confere
The multi-stage search method shown in "ncing" has been proposed.
The two-step search, which is an example of the multi-step search, will be described below.

As shown in FIG. 9, in the first-stage search, the search density is rough because the search points are extracted by skipping the pixels in the horizontal scanning direction and the vertical scanning direction on the reference image. However, a wide search area is set, and the motion vector MV1 of the first stage is selected from the set search points.
Is required. In this first stage, the search is made approximately where the motion vector is. Next, in the search of the second stage, a search density that is narrower than the search region of the first stage and extracts all the pixels in the region near the search point corresponding to the motion vector MV1 obtained in the first stage. Is set, and the second-stage motion vector MV2 is obtained from this search area. Next, the motion vector MV1 obtained in the first stage and the MV2 obtained in the second stage are added to obtain the final motion vector MV.

As described above, since the search points can be reduced by searching the motion vector in multiple stages from the coarse search to the fine search, the calculation amount and the memory access amount for searching the motion vector MV. While reducing
A motion vector with high prediction accuracy is efficiently obtained. Further, as a method for improving the processing capability for searching a motion vector, for example, the Institute of Electronics, Information and Communication Engineers, 1989 Vo.
l. J72-D-2 No. No. 3, page 395 to page 402, "hierarchical search method shown in" Motion amount detection method in moving image using hierarchical pixel information ". In this method, the average pixel value of the input image is calculated and a reduced image is generated to generate an upper layer with degraded spatial resolution, and the detection result of the upper layer (coarse search) that detects the global movement of the subject By successively detecting the amount of motion of the lower layer (fine search) where the resolution becomes finer, the number of search points is substantially reduced, and the motion vector is efficiently and highly accurately obtained.

[0021]

However, the frame motion vector MVA, the first field motion vector MVA1 and the second field motion vector MVA described above are used.
2, and the field motion vector MVB, the first segment motion vector MVB1 and the second segment MV
In the conventional moving image compression method and apparatus, such as B, each of a plurality of motion vectors is divided into a coarse search and a fine search, and an optimum motion vector is determined from the searched plurality of motion vectors. Since the optimum motion vector was determined from among the motion vectors finally obtained through the finest search, the difference in the amount of processing for searching the plurality of motion vectors becomes large according to the picture type, There is a problem that the encoding processing speed varies.

That is, the P picture uses the temporally forward image as the reference image, while the B picture uses the temporally forward and backward images as reference images. Therefore, in the case of the B picture, However, there is a problem in that a larger amount of calculation and a larger memory access amount (bandwidth) are required than in the case of the P picture, and the motion vector search processing of the B picture hinders the improvement of the overall coding speed.

Therefore, according to the present invention, in the moving image compression method and apparatus for dividing the coarse search into the fine search, searching each of the plurality of motion vectors to determine the optimum motion vector, the optimum motion is determined. By performing the motion vector determination based on the search result in the middle according to the picture type, for example,
In the case of a P picture, motion vector determination is performed from the motion vector obtained through the final fine search, and B
In the case of a picture, the motion vector determination is performed from the motion vector obtained in the middle search, so that the motion vector search processing amounts of the P picture and the B picture are averaged to efficiently perform the motion vector search processing. It is an object of the present invention to provide a moving image compression method and device capable of speeding up the encoding process without degrading the image quality of.

[0024]

In order to solve the above-mentioned problem, the invention as set forth in claim 1 relates to each frame forming a moving image of an interlaced scanning system, and a plurality of pixels included in pixels forming the frame. Each type of image information, including a format conversion step of extracting in the horizontal scanning direction and the vertical scanning direction based on a predetermined sampling frequency, for each frame extracted by the format conversion step,
An intra-frame prediction image coded based on intra-frame information, a forward prediction image coded using a temporally forward frame as a reference image, and temporally forward and backward frames are referred to. A bidirectional predictive image encoded as an image, and a moving image compression method for switching the generated images, wherein the intraframe predictive image, the forward predictive image, and the bidirectional predictive image are encoded. An image discrimination step of discriminating a predicted image, the frame including a first field and a second field, and a current image frame block partially forming the frame includes a current image first field block including the first field. A current image frame block including the current image second field block including the second field, a current image first field A first search area setting step of setting a wide search area with a coarse search density by extracting image information representative of the pixels forming the reference image for the lock and the second field block of the current image And a similar block that is most similar to the current image frame block, the current image first field block and the current image second field block is specified based on the search region set by the first search region setting step, and the current image is determined. The frame motion vector corresponding to the frame block, the first corresponding to the current image first field block
A first motion vector searching step of searching for a second field motion vector corresponding to the field motion vector and the current image second field block; and a current image frame block specified by the first motion vector searching step,
In the vicinity of the similar blocks of the current image first field block and the current image second field block,
A second search area setting step of setting a search area having a finer search density and a narrower search area than the search area set by the first search area setting step, and a second search area setting step set by the second search area setting step, respectively. Based on the search area of the current image frame block, the current image first field block and the current image second field block, the similar blocks are respectively identified, and the frame motion vector corresponding to the current image frame block and the current image first block are identified. 1
A second motion vector searching step of searching for a first field motion vector corresponding to a field block and a second field motion vector corresponding to a current image second field block; and the above-mentioned search by the first motion vector searching step. A first motion vector determining step of determining an optimum motion vector from among a frame motion vector, a first field motion vector and a second field motion vector; and the frame motion vector searched by the second motion vector searching step , A second motion vector determination step of determining an optimum motion vector from the first field motion vector and the second field motion vector, and a first motion vector according to the predicted image determined by the image determination step.
A selection step of selecting one of the motion vector determination step and the second motion vector determination step;
It is characterized by including.

According to a second aspect of the present invention, in the first aspect of the invention, when the predicted image discriminated by the image discriminating step is a forward predictive image, the selecting step is the second motion vector determining step. And when the predicted image determined by the image determination step is a bidirectional predicted image, the selection step selects the first motion vector determination step.

According to a third aspect of the present invention, for each field forming a moving image of the interlaced scanning system, a plurality of types of image information contained in the pixels forming the field are respectively set to a predetermined sampling frequency. Based on the intra-field prediction image encoded based on the intra-field information, and temporally including a format conversion step of extracting in the horizontal scanning direction and the vertical scanning direction based on each field extracted by the format conversion step. A moving image generated by switching between a forward prediction image that is encoded using the forward field as a reference image and a bidirectional prediction image that is temporally encoded using the forward and backward fields as reference images. An image compression method, comprising: the intra-field predicted image, the forward predicted image, and the bidirectional predicted image. An image discrimination step of discriminating a predicted image to be coded, a current image field block partially forming the field, and a current image first segment block consisting of an upper half block of the current image field block and the current image An image including a second segment block of the current image composed of lower half blocks of the field block, and an image representing a pixel forming the reference image for the current image field block, the first segment block of the current image and the second segment block of the current image A first search area setting step of setting a wide search area having a coarse search density by extracting information, and the current image field based on the search area set by the first search area setting step. Block, current image first segment block and current image second segment block Tsu identify similar blocks most similar respective click, field motion vector corresponding to the current image field block, the current picture first
A first motion vector searching step of searching for a first segment motion vector corresponding to a segment block and a second image corresponding to a current image second segment block; and a current motion vector specified by the first motion vector searching step. Near the similar blocks of the image field block, the first segment block of the current image and the second segment block of the current image, the search density is finer than the search area set by the first search area setting step, and narrow. A second search area setting step of setting respective search areas, and a current image field block, a current image first segment block and a current image second based on the respective search areas set by the second search area setting step. Each similar block that is the most similar to the segment block And a second motion vector searching for a field motion vector corresponding to the current image field block, a first segment motion vector corresponding to the current image first segment block and a second segment motion vector corresponding to the current image second segment block A search step, and a first motion vector determination step of determining an optimum motion vector from the field motion vector, the first segment motion vector, and the second segment motion vector searched by the first motion vector search step. , The field motion vector searched by the second motion vector searching step, the first segment motion vector,
A second motion vector determination step of determining an optimum motion vector from the second segment motion vectors, and a first motion vector determination step and a second motion vector according to the predicted image determined by the image determination step. And a selection step of selecting one of the determination steps.

According to a fourth aspect of the present invention, in the invention according to the third aspect, when the predicted image determined by the image determining step is a forward predictive image, the selecting step is the second motion vector determining step. And when the predicted image determined by the image determining step is a bidirectional predicted image, the selecting step selects the first motion vector determining step. According to a fifth aspect of the present invention, for each frame forming a moving image of the interlaced scanning system, a plurality of types of image information included in pixels forming the frame are horizontally transmitted based on a predetermined sampling frequency. A format conversion unit for extracting in the scanning direction and the vertical scanning direction is provided, and for each frame extracted by the format conversion unit, an intra-frame predicted image coded based on intra-frame information and a temporal forward direction. Moving image compression generated by switching between a forward predictive image coded as a reference image of each frame and a bidirectional predictive image encoded as a temporally forward and backward frame as a reference image. In a device, a predicted image to be encoded is determined from the intra-frame predicted image, the forward predicted image, and the bidirectional predicted image. Image discriminating means, the frame comprises a first field and a second field, and a current image frame block which partially constitutes the frame is a current image first field block and the second field which are the first field. By extracting image information representative of pixels forming the reference image for the current image frame block, the current image first field block and the current image second field block, First search area setting means that sets a wide search area having a coarse search density, and the current image frame block and the current image first field based on the search areas set by the first search area setting means Block and the similar block that is most similar to the second field block of the current image. A frame motion vector corresponding to the current image frame block, a first field motion vector corresponding to the current image first field block, and a second field motion vector corresponding to the current image second field block. One motion vector searching means, and the first image vector block, the current image first field block, and the current image second field block specified by the first motion vector searching means, in the vicinity of the similar blocks, respectively. A second search area setting unit that sets a narrower search area and has a finer search density than the search area set by the first search area setting unit, and each search set by the second search area setting unit. The current image frame block based on region,
The similar blocks most similar to the current image first field block and the current image second field block are respectively specified, and the frame motion vector corresponding to the current image frame block, the first field motion vector corresponding to the current image first field block, and Second motion vector searching means for searching a second field motion vector corresponding to the current image second field block; the frame motion vector searched by the first motion vector searching means; the first field motion vector; First to determine the optimum motion vector from the two-field motion vector
Second motion vector determining means, and a second motion vector determining an optimum motion vector from the frame motion vector, the first field motion vector, and the second field motion vector searched by the second motion vector searching means. And a selection unit for selecting one of the first motion vector determination unit and the second motion vector determination unit according to the predicted image determined by the image determination unit. It is characterized by that.

According to a sixth aspect of the present invention, in the invention according to the fifth aspect, when the predicted image discriminated by the image discriminating means is a forward predictive image, the selecting means is the second motion vector discriminating means. And when the prediction image determined by the image determination means is a bidirectional prediction image, the selection means selects the first motion vector determination means.

According to a seventh aspect of the present invention, for each field forming a moving image of the interlaced scanning system, a plurality of types of image information contained in pixels forming the field are respectively set to predetermined sampling frequencies. A format conversion means for extracting in the horizontal scanning direction and the vertical scanning direction on the basis of each of the fields extracted by the format conversion means; A forward prediction image that is encoded using the forward field as a reference image, and a bidirectional prediction image that is temporally encoded using the forward and backward fields as reference images. In the moving picture compression device, among the intra-frame predicted image, the forward predicted image, and the bidirectional predicted image, An image discriminating means for discriminating a predicted image to be encoded, a current image field block partially forming the field, and a current image first segment block consisting of an upper half block of the current image field block and the current image An image including a second segment block of the current image composed of lower half blocks of field blocks, and an image representing a pixel forming the reference image for the current image field block, the first segment block of the current image and the second segment block of the current image First search area setting means for respectively setting a wide search area having a coarse search density by extracting information, and the current image field based on the search area set by the first search area setting means. Block, current image first segment block and current image second segment block Identifying each the most similar similar blocks click, field motion vector corresponding to the current image field block, a second corresponding to the first segment motion vectors and the current picture second segment blocks corresponding to the first segment block current picture
First motion vector searching means for searching a segment motion vector, and the similarity of each of the current image field block, the current image first segment block and the current image second segment block specified by the first motion vector searching means. A second search area setting means and a second search area setting means which respectively set a search area having a finer search density and a narrower search area than the search area set by the first search area setting means in the vicinity of the block. Based on the respective search areas set by, the similar blocks most similar to the current image field block, the current image first segment block and the current image second segment block are respectively specified, and the field motion corresponding to the current image field block is specified. Vector, the first corresponding to the first segment block of the current image A second motion vector search means for searching the segments motion vectors and the second segment motion vector corresponding to the current image second segment blocks,
First motion vector determination means for determining an optimum motion vector from the field motion vector, first segment motion vector, and second segment motion vector searched by the first motion vector search means; Discriminating by the image discriminating means and the second motion vector discriminating means for discriminating the optimum motion vector from the field motion vector, the first segment motion vector and the second segment motion vector searched by the motion vector searching means It is characterized by further comprising: a selection unit that selects one of the first motion vector determination unit and the second motion vector determination unit according to the predicted image.

According to an eighth aspect of the invention, in the seventh aspect of the invention, when the predicted image discriminated by the image discriminating means is a forward predictive image, the selecting means is the second motion vector determining means. And when the predicted image determined by the image determination means is a bidirectional predicted image, the selection means selects the first motion vector determination means.

[0031]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below with reference to the drawings. 1 and 2 are diagrams showing a moving image compression apparatus according to an embodiment of the present invention. As shown in FIG. 1, this moving picture compression apparatus includes a preprocessing unit 11, a coding mode control unit 12, a prediction error signal generation unit 13, and a DCT (Discrete Cosine Transfor).
m; Discrete Cosine Transform) unit 14, quantization unit 15, variable length coding unit 16, buffer unit 1
7, code amount control unit 18, dequantization unit 21,
The inverse DCT unit 22, the reproduced image generation unit 23, the reproduced image storage unit 24, the motion compensation prediction unit 25, and the motion vector search unit 26 are provided, and each unit is connected to each other by a bus (not shown).

Reference numeral 1 denotes an image signal input terminal for inputting a digital moving image signal from an image memory (not shown). Through the image signal input terminal 1, each screen constituting a moving image is input to the preprocessing unit 11 in the order of encoding. To do. The image signal is data including luminance information and color difference information corresponding to the pixels forming each screen. The pre-processing unit 11 has a function of a format conversion means, and based on a predetermined sampling frequency, the luminance information and the color difference information contained in the pixels forming the screen are applied to the image signal input through the image signal input terminal 1. Then, it is extracted in the horizontal scanning direction and the vertical scanning direction, and is converted into a so-called 4: 2: 0, 4: 2: 2 or 4: 4: 4 image format. In addition, the pretreatment unit 11
Divide the format-converted image signal of each screen into rectangular blocks of 16 pixels × 16 pixels (hereinafter referred to as macroblocks), and the prediction error signal generation unit 13 and the motion vector search unit 26 for each of these macroblocks. Output to.

The coding mode control unit 12 has a function of image discrimination means, and based on the header information contained in each screen input to the preprocessing unit 11, the screen structure (frame structure or field structure) of each screen. ) And a picture type (I picture, P picture or B picture) are discriminated, and the discriminated screen structure and picture type are respectively referred to as a prediction error signal generation unit 13, a reproduced image storage unit 24, a motion compensation prediction unit 25 and a motion vector. It outputs to the search unit 26 and the variable length coding unit 16.

Further, the coding mode control unit 12 is
Determines each macroblock attribute of each macroblock in the picture according to the determined picture type,
Output in the same way. Macroblock attributes include an intra block that simply encodes an input image, a simple inter-frame block that encodes a difference image without motion compensation, and a difference image that performs motion compensation. There is a motion-compensated inter-frame block to be converted.

If the determined picture type is an I picture, all macro blocks in the picture are intra blocks. On the other hand, when the determined picture type is a P picture or a B picture, when an intra slice is designated in advance in the header information, all macroblocks in the slice are treated as intra blocks, and in other cases, motion is performed. The estimated value of the generated code amount of the image information of the macroblock in each macroblock attribute such as the intra block, the simple inter-frame block, the motion-compensated inter-frame block output from the compensation unit 25, and the motion vector search unit 26
Based on the motion vector information output from the macroblock, the macroblock attribute that minimizes the generated code amount of the macroblock is determined.

The prediction error signal generation unit 13 is based on the macroblock attribute output from the coding mode control unit 12, and is an input image (intra block), and a difference image between the input image and a prediction image for which motion compensation is not performed Any one of (a simple inter-frame block), a difference image between the input image and the motion-compensated predicted image (motion-compensated inter-frame block), etc. is output to the DCT unit 14 as a prediction error signal.

The DCT unit 14 divides the image signal of the macro block output from the prediction error signal generation unit 13 into blocks of 8 pixels × 8 pixels, performs a known two-dimensional DCT operation for each block, and outputs the image data. Is converted into a plurality of DCT coefficients from a low frequency term to a high frequency term including a DC component (DC component) and an AC component (AC component).

By this DCT operation, the image data of each block is decomposed into a plurality of stepwise frequency components that gradually express fineness, from the DC component which is the first low frequency term to the AC component of the high frequency term. . Before the DCT operation,
Since the pixel values (luminance information and color difference information) of the image signal exhibiting a low spatial correlation are concentrated on the low-frequency terms by the DCT operation, effective information compression can be performed by removing the high-frequency terms.

Although DCT is adopted as a preferable orthogonal transform method, it is not limited to this. For example, current television signals have a mixture of component signals and composite signals, and DCT shows excellent characteristics for component signals, but Hadamard transform shows excellent characteristics for composite signals. Are known. As described above, an orthogonal transform method suitable for the characteristics of the image to be encoded may be adopted.

The quantization unit 15 is the DCT unit 1
The DCT coefficient output from No. 4 is divided by a quantization step, the remainder is rounded to remove high frequencies, and quantization is performed, and output to the variable length coding unit 16 and the inverse quantization unit 21. D obtained by DCT operation
The DC component and the AC component of the CT coefficient are quantized by independent quantization steps, and the DC of the intra block is quantized.
For the coefficient, the quantization step is controlled on a picture-by-picture basis, and for the other coefficients, the quantization step is controlled by multiplying the quantization matrix according to the picture type by the quantization scale determined for each macroblock. The quantization matrix is a parameter for adjusting the quantization characteristics to the visual characteristics. The quantization scale is a parameter for controlling the generated code amount, and is controlled by the code amount control unit 18 described later.

The variable-length coding unit 16 scans the DCT coefficients quantized by the quantization unit 15 in order from the low frequency component, thereby converting the two-dimensional coefficients into a one-dimensional coefficient sequence, and Is subjected to variable length coding on the basis of the run representing the number of consecutive zero coefficients of the one-dimensional coefficient sequence and the level representing the value of the non-zero coefficient, and the encoded data is output to the buffer unit 17. The scan order is
There are so-called zig-zag scanning and so-called alternate scanning, which are switched and used according to the type of image. The alternate scan is effective for an interlaced image. Also, the variable length coding unit 16 performs variable length coding on the motion vector information output from the motion vector search unit 26, and outputs the motion vector information to the buffer unit 17.

The buffer unit 17 inputs the encoded data from the variable length encoding unit 16 and outputs it while averaging the bit rates. That is, since the generated code amount of the buffer unit 17 varies depending on the complexity of the image and the intensity of the motion, the buffer unit 17 absorbs the variation and transmits the encoded data as a bit stream at a substantially constant transmission rate. The encoded data output from the buffer unit 17 is output through the output terminal 2 and transmitted to the external decoder via a transmission line (not shown).

The code amount control unit 18 grasps the generated code amount by monitoring the occupation rate of the coded data in the buffer unit 17, and controls the quantization scale of the quantization unit 15 so as to match the target bit rate. The inverse quantization unit 21 performs inverse quantization by multiplying each DCT coefficient quantized by the quantization unit 15 by the same quantization step as in the case of quantization, and the inverse DCT unit 2
Output to 2. The inverse DCT unit 22 performs an inverse DCT operation on the inversely quantized two-dimensional coefficient, and then synthesizes a macroblock of 16 pixels × 16 pixels from each block of 8 pixels × 8 pixels to obtain a restored image of the prediction error signal. Is generated and output to the reproduction image generation unit 23.

The reproduced image generation unit 23 generates a reproduced image corresponding to the original image based on the restored image of the prediction error signal output from the inverse DCT unit 22 and the predicted image output from the motion compensation prediction unit 25, It is stored in the reproduced image storage unit 24. That is, the inverse quantization unit 21, the inverse DCT unit 22, and the reproduction image generation unit 23 perform local decoding, and the reproduction image stored in the reproduction image storage unit 24 is a P picture or a P picture to be encoded later. It becomes the reference image of the B picture.

The motion compensation prediction unit 25, based on the reproduced image stored in the reproduced image generation unit 24 and the motion vector searched by the motion vector search unit 26, intra block, simple inter-frame block, motion compensated inter-frame block. A predictive image in each macroblock attribute such as, for example, is generated, the generated code amount is estimated when the predictive error signal is output from the predictive error signal generating unit 13, and the encoding mode control unit 1
Output to 2.

In the coding mode control unit 12, the estimated value of the generated code amount when the prediction error signal in each macroblock attribute outputted by the motion compensation prediction unit 25 and the motion vector search unit 26 output it. Based on the motion vector information, the macroblock attribute that minimizes the generated code amount of the macroblock is determined. The motion compensation prediction unit 25 outputs the predicted image to the reproduced image generation unit 23 based on the macroblock attribute determined by the coding mode control unit 12.

The motion vector search unit 26 receives an image of the input image (current image) for each macroblock output from the preprocessing unit 11 according to the screen structure and the picture type output from the coding mode control unit 12. A motion vector is obtained based on the information and the image information of the reference image (I picture or P picture already encoded) stored in the reproduced image storage unit 24, and the motion vector information is obtained by the encoding mode control unit 12 and the motion compensation prediction. The data is output to the unit 26, the reproduced image storage unit 24, and the variable length coding unit 16.

Here, when the screen structure is a frame structure, the 16 pixel × 16 pixel macro block corresponds to the current image frame block described in the prior art.
The motion vector search unit 26 includes a frame motion vector MVA corresponding to the current image frame block, a first field motion vector MVA1 corresponding to the current image first field block and a second field motion vector MVA2 corresponding to the current image second field block. From each of these motion vectors, an optimum one frame motion vector MVA or two field motion vectors MVA1 and MVA2 are selected and output.

On the other hand, when the screen structure is the field structure, the macroblock of 16 pixels × 16 pixels corresponds to the current image field block described in the prior art, and the motion vector search unit 26 determines the current image field block. Field motion vector MVB, a first segment motion vector MVB1 corresponding to the current image first segment block, and a second segment motion vector MVB2 corresponding to the current image second segment block, respectively. The optimum one field motion vector MVB or two segment motion vectors MVB1 and MVB2 are selected and output.

Further, the motion vector search unit 26 is
Rough search (first stage search) and fine search (second stage search) shown in FIG.
2, the determination stage switching unit 30, the first search area setting unit 31, the first motion vector search unit 32, the second motion vector search unit 32, and the second motion vector search unit 32.
A search area setting unit 33, a second motion vector search unit 34, a motion vector addition unit 35, and a motion vector determination unit 36 are provided. Each unit is
They are connected to each other by a local bus 37 connected to the bus.

The judgment step switching unit 30 has a function of selecting means, switches the step of judging the motion vector based on the picture type output from the coding mode control unit 12, and outputs information representing the judgment step to the first step. 1 motion vector search unit 32, second search area setting unit 3
3 to the second motion vector search unit 34 and the motion vector determination unit 36.

Here, in the judging step, the picture type is P
In the case of a picture, it represents the second stage, and when the picture type is a B picture, it represents the first stage. That is, in the case of a P picture, an optimum motion vector is obtained from the three motion vectors obtained by the first-stage search and the second-stage search for the three types of current image blocks included in the macroblock. On the other hand, in the case of the B picture, the optimum motion vector is determined from the three motion vectors obtained by the search in the first stage.

The first search area setting unit 31 sets the search areas corresponding to the three types of current image blocks included in the macroblock based on the screen structure output from the coding mode control unit 12. Each search area has a wide search area although the search density is coarse by extracting the search points from the reference image stored in the reproduced image storage unit 24 by skipping pixels in the horizontal scanning direction and the pixels in the vertical scanning direction. Is set.

The first motion vector search unit 32, for each macroblock output from the preprocessing unit 11,
Based on the screen structure output from the coding mode control unit 12, a motion vector is selected from each of the search areas set by the first search area setting unit 31 for the three types of current image blocks included in the macroblock. Then, the first-stage motion vector search with rough prediction accuracy is performed.

Specifically, first, the distortion representing the difference between the pixel data of each candidate block and the pixel data of the current image block is calculated. Then, the candidate block (similar block) most similar to the current image block is specified by specifying the distortion having the minimum value (minimum distortion) from the calculated distortions. Then, the displacement between the similar block and the current image block is calculated to obtain the motion vector.

The obtained motion vector and minimum distortion are output based on the information indicating the judgment stage output from the judgment stage switching unit 30. here,
When the determination step is the first step (B picture), all the obtained motion vectors and the minimum distortion are output to the motion vector determination unit 36. On the other hand, when the determination step is the second step (P picture), all the obtained motion vectors are the second search area setting unit 3
3 to the second motion vector search unit 34 and the motion vector addition unit 35.

The second search area setting unit 33 inputs the motion vector of the first stage from the first motion vector searching unit 32 or the motion vector judging unit 36, and the current image block of the input motion vector of the first stage is inputted. The search area is set near the candidate block (similar block) searched by the first motion vector search unit 32. These search areas are narrower than the search area set by the first search area setting unit 31, and the image information of all the pixels in the search area is the reproduced image storage unit 24.
A search area with a fine search density is set so that the search area is extracted from

The second motion vector search unit 34 inputs the first stage motion vector from the first motion vector search unit 32 or the motion vector determination unit 36,
For the current image block corresponding to the input first-step motion vector, the motion vector and the minimum distortion are obtained from the search area set by the second search area setting unit 33, and the second step with a fine prediction accuracy is obtained. Motion vector search is performed.

Here, the motion vector and the minimum distortion are obtained in the same manner as in the first motion vector search unit 32, but the motion vector is searched for the similar block specified by the first motion vector search unit 32 and the second motion vector search. The displacement with the similar block identified by the unit 34 is represented. The calculated motion vector and minimum distortion are calculated by the motion vector addition unit 35.
Is output to

The motion vector addition unit 35 adds the first stage motion vector and the second stage motion vector corresponding to the same current image block to obtain the final motion vector of each current image block. The final motion vector is output based on the information indicating the determination stage output from the determination stage switching unit 30. Here, when the determination step is the first step (B picture), the final motion vector is output as it is to the coding mode control unit 12, the motion compensation prediction unit 25, and the like. On the other hand, when the determination stage is the second stage (P picture), the final motion vector is output to the motion vector determination unit 36.

The motion vector determination unit 36 inputs the first stage motion vector output from the first motion vector search unit 32 or the final motion vector output from the motion vector addition unit 35, and outputs the motion vector of the input motion vector. The optimum motion vector is selected from the output and output. Here, when the screen structure is a frame structure, first, the minimum distortion of the first field motion vector and the minimum distortion of the second field motion vector are added, and then the added minimum distortion and the frame motion vector are added. The minimum distortion is compared, and the smaller value is determined as the optimum motion vector.

On the other hand, when the screen structure is the field structure, first, the minimum distortion of the first segment motion vector and the minimum distortion of the second segment motion vector are added, and then the added minimum distortion and the field motion are added. The minimum distortion of the vector is compared, and the smaller value is determined as the optimum motion vector.

These motion vectors are output based on the information indicating the determination stage output from the determination stage switching unit 30. Here, when the determination step is the first step (B picture), the selected motion vector is output to the second search area setting unit 33 and the second motion vector search unit 34. On the other hand, when the determination step is the second step (P picture), the selected motion vector is output to the coding mode control unit 12, the motion compensation prediction unit 25, and the like.

That is, in the present moving picture compression apparatus, the motion vector search unit 26 previously determines the optimum motion vector from one frame motion vector and two field motion vectors in the case of a frame structure. In addition, in the case of the field structure, the optimum motion vector is determined from one field motion vector and two segment motion vectors. Then, the determination stage switching unit 30 switches the timing of determining the motion vector according to the picture type. When the picture type is the P picture, the first and second
The optimum motion vector is determined from the final motion vectors obtained by adding the motion vectors of the steps, while the first step obtained by the first motion vector search unit 32 when the picture type is B picture. The optimum motion vector is determined from the motion vector of

The motion vector of the first stage has a coarse search density but is obtained from a wide search area, and therefore the size of the motion vector is equal to or several times as large as that of the motion vector of the second stage. . On the other hand, the motion vector of the second stage has a small search density but is obtained from a narrow search region, and thus the size of the motion vector is small. Therefore, in the case of a B picture, a slight deterioration in image quality is allowed by making a motion vector determination based on the first stage motion vector. In case of B picture,
This is because even if the image quality is lowered to some extent, it does not significantly affect the image quality deterioration of the entire moving image. However, in the case of a P picture, if the image quality is deteriorated, it will affect the image quality deterioration of the entire moving image. Therefore, in the case of the P picture, the optimum motion vector is determined from the final motion vector added by the motion vector addition unit 35.

Next, an embodiment of the present invention will be described. Hereinafter, the operation of searching the motion vector by the motion vector search unit 26 for each screen will be described based on the flowchart shown in FIG. Here, the screen structure and picture type of each screen determined by the coding control mode control unit 12 are input to the motion vector search unit 26, and in the case of P picture or B picture, the intra slice is designated. It is assumed that the macroblock attribute previously determined as an intra block is input from the coding mode control unit 12 to the motion vector search unit 26.

First, the picture type output from the coding mode control unit 12 is determined (step S
1). Here, when it is determined that the picture type is an I picture, all macroblocks are intra blocks, and therefore the motion vector search is not performed and the process for the screen ends. When it is determined in step S1 that the picture type is the P picture, the determination step switching unit 30 switches the determination step to the second step (step S2).

Next, it is judged whether or not there is a macroblock to be processed (step S3). Here, if there is no macroblock to be processed, the process for that screen is terminated. On the other hand, when there is a macroblock to be processed, it is determined whether or not the macroblock is determined in advance as an intra block (step S4).

If the macroblock attribute is determined to be an intrablock, the motion vector search is not performed and the process returns to step S3. On the other hand, when the macroblock attribute is not determined as the intra block, the first search area setting unit 31 sets the search area corresponding to the current image block included in the macroblock, and the pixel data of the set search area is set. Based on the pixel data of the macroblock and the macroblock, the first motion vector search unit 42 searches for the first-stage motion vector, and outputs it to the second search area setting unit 33 and the second motion vector search unit 34 (step S5). .

Next, the first motion vector search unit 32
The search area is set by the second search area setting unit 42 with respect to the motion vector of the first stage searched by the second motion vector based on the pixel data of the set search area and the pixel data of the macro block. The search unit 34 searches for the second stage motion vector and outputs it to the motion vector addition unit 35 (step S
6).

Next, the motion vector addition unit 35 adds the first stage motion vector and the second stage motion vector corresponding to the same current image block, and the final added motion vector is the motion vector determination unit. 36
(Step S7). Next, the motion vector determination unit 36 selects the optimum motion vector from the final motion vectors, and the motion compensation prediction unit 25
To step S3 (step S3).
8).

Further, when it is determined in step S1 that the picture type is the B picture, the determination step switching unit 30 switches the determination step to the first step (step S9). Next, it is determined whether there is a macroblock to be processed (step S10). Here, if there is no macroblock to be processed, the process for that screen is terminated. On the other hand, if there is a macroblock to be processed, it is determined whether or not the macroblock has been previously determined as an intra block (step S11).

If the macroblock attribute is determined to be an intra block, the motion vector search is not performed and the process returns to step S10. On the other hand, when the macroblock attribute is not determined as the intra block, the first search area setting unit 31 sets the search area corresponding to the current image block included in the macroblock, and the pixel data of the set search area is set. And the first motion vector search unit 42 searches the first stage motion vector based on the pixel data of the macroblock,
It is output to the motion vector determination unit 36 (step S12).

Next, the first motion vector search unit 32
The optimum motion vector is selected by the motion vector determination unit 36 from the first-stage motion vectors searched for by and is output to the second search area setting unit 33 and the second motion vector search unit 34, respectively (step S13). ). Next, the motion vector determination unit 36
To the first stage motion vector selected by
The search area is set by the search area setting unit 33,
The second motion vector search unit 34 based on the pixel data of the set search area and the pixel data of the macro block.
Thus, the second stage motion vector is searched for and output to the motion vector addition unit 35 (step S14).

Next, the motion vector addition unit 35 adds the first-stage motion vector and the second-stage motion vector corresponding to the same current image block, outputs the result to the motion compensation prediction unit 25, and again performs the step. It returns to S10 (step S15). In this way, since the motion vector is searched by switching the determination stage according to the picture type output from the coding mode control unit 12, in the case of the P picture, the motion vectors of the first and second stages are The optimum motion vector is determined based on the added final motion vector, and in the case of the B picture, the optimum motion vector is determined based on the first stage motion vector.

Therefore, the motion vector search processing amount of the B picture can be significantly reduced, so that the motion vector search processing amount of the P picture and the B picture can be averaged to efficiently perform the motion vector search processing. .
Therefore, the encoding process can be speeded up without degrading the image quality of the entire moving image. In the motion vector search 26 of the moving picture compression apparatus according to the present embodiment, a multi-step search method is adopted, and the timing of determining the motion vector is switched according to the picture type. It is needless to say that even when the motion vector of the upper layer is searched for by the method, the motion vector search process can be efficiently performed by similarly switching the determination time according to the picture type.

[0077]

According to the first aspect of the present invention, the first image is detected according to the type of the predicted image discriminated in the image discriminating step.
Since it is possible to select either one of the second and second determination steps, in the case of a predicted image in which the amount of motion vector search processing is large and the decrease in motion vector prediction accuracy is acceptable, By determining the optimum motion vector in the determination step of 1, it is possible to reduce the motion vector search processing amount and average the motion vector search processing amount of each prediction image. Therefore, the encoding process can be speeded up without degrading the image quality of the entire moving image.

According to the invention of claim 2, claim 1
In the invention described in (3), when the predicted image is the forward predicted image, the optimal motion vector is determined in the second motion vector determination step, and when the predicted image is the bidirectional predicted image, the first motion vector determination step Since the optimum motion vector is determined by, it is possible to reduce the motion vector search processing amount of the bidirectional predicted image and average the motion vector search processing amount of the forward predicted image and the bidirectional predicted image.
Therefore, the encoding process can be speeded up without degrading the image quality of the entire moving image.

According to the third aspect of the present invention, one of the first and second determination steps can be selected according to the type of the predicted image determined by the image determination step. Therefore, in the case of a predicted image in which the amount of motion vector search processing is large and the prediction accuracy of the motion vector can be reduced, the motion vector search process is performed by determining the optimum motion vector in the first determination step. By reducing the amount, the motion vector search processing amount of each prediction image can be averaged. Therefore, the encoding process can be speeded up without degrading the image quality of the entire moving image.

According to the invention of claim 4, claim 3
In the invention described in (1), when the predicted image is a forward predicted image, the optimal motion vector is determined in the second motion vector determination step, and when the predicted image is a bidirectional predicted image, the first motion vector determination step is performed. Since the optimum motion vector is determined, it is possible to reduce the motion vector search processing of the bidirectional predicted image and average the motion vector search processing amounts of the forward predicted image and the bidirectional predicted image. Therefore, without degrading the image quality of the entire moving image,
The encoding process can be speeded up.

According to the fifth aspect of the invention, one of the first and second determining means can be selected according to the type of the predicted image determined by the image determining means. Therefore, in the case of a predicted image in which the amount of motion vector search processing is large and the prediction accuracy of the motion vector can be reduced, the motion vector search processing is performed by determining the optimum motion vector by the first determination unit. The amount can be reduced and the amount of processing for searching the motion vector of each predicted image can be averaged. Therefore, the encoding process can be speeded up without degrading the image quality of the entire moving image.

According to the invention described in claim 6, claim 5
In the invention described in (1), when the predicted image is a forward predicted image, the second motion vector determination means determines the optimum motion vector, and when the predicted image is a bidirectional predicted image, the first motion vector determination means Since the optimum motion vector is determined, it is possible to reduce the motion vector search processing amount of the bidirectional predicted image and average the motion vector search processing amount of the forward predicted image and the bidirectional predicted image.
Therefore, the encoding process can be speeded up without degrading the image quality of the entire moving image.

According to the invention described in claim 7, either one of the first and second determining means can be selected according to the type of the predicted image determined by the image determining means. Therefore, in the case of a predicted image in which the motion vector search processing amount is large and the prediction accuracy of the motion vector can be reduced, the motion vector search process is performed by determining the optimum motion vector by the first determination unit. The amount can be reduced and the amount of processing for searching the motion vector of each predicted image can be averaged. Therefore, the encoding process can be speeded up without degrading the image quality of the entire moving image. According to the invention described in claim 8, in the invention described in claim 7, when the predicted image is a forward predicted image, the second motion vector determination means determines the optimum motion vector and In the case of a bidirectionally predicted image, the first motion vector determination means determines the optimal motion vector, so the amount of processing for searching the motion vector of the bidirectionally predicted image is reduced, and the motions of the forward and backward predicted images are calculated. The vector search processing amount can be averaged. Therefore, the encoding process can be speeded up without degrading the image quality of the entire moving image.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a moving image compression apparatus according to an embodiment of the present invention.

2 is a block diagram showing a configuration of a motion vector search unit 26. FIG.

FIG. 3 is a flowchart showing an operation of searching a motion vector by a motion vector search unit 26.

FIG. 4 is a diagram showing a picture structure in MPEG.

FIG. 5 is a diagram showing a moving image having a frame structure.

FIG. 6 is a diagram showing motion vectors of a frame prediction method and a field prediction method in a frame structure.

FIG. 7 is a diagram showing a moving image of a field structure.

[Fig. 8] Fig. 8 is a diagram illustrating motion vectors of a 16x16 field prediction method and a 16x8 field prediction method in a field structure.

FIG. 9 is a diagram showing a motion vector search method based on a two-step search method including a coarse search and a fine search.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Image signal input terminal 2 Bit stream output terminal 11 Pre-processing unit (format conversion means) 12 Coding mode control unit (image discrimination means) 13 Prediction error signal generation unit 14 DCT unit 15 Quantization unit 16 Variable length coding unit 17 Buffer unit 18 Code amount control unit 21 Inverse quantization unit 22 Inverse DCT unit 23 Reproduced image generation unit 24 Reproduced image storage unit 25 Motion compensation prediction unit 26 Motion vector search unit 30 Judgment stage switching unit (selection means) 31 First search area Setting unit (first search area setting means) 32 First motion vector search unit (first motion vector searching means) 33 Second search area setting unit (second search area setting means) 34 Second motion vector searching unit Second motion vector search means) 35 Motion vector addition unit 36 Motion vector determination unit (first and second motion vector determination means) 37 Local bus 110 Current image frame 111 Current image first field 112 Current image Second field 120 Reference image frame 121 Reference image first field 122 Reference image second field 210 Current image frame block 211 Current image first field block 212 Current image second field block 220 Reference image frame candidate block 221 Reference image first field candidate block 222 Reference Image second field candidate block 310 Current image field 320 Reference image field 410 Current image field block 211 Current image first segment block 212 Current image second segment block Click 420 reference picture field candidate block 421 the reference image first segment candidate block 422 the reference image second segment candidate block

 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Norihiko Nagai 4-36-19 Yoyogi, Shibuya-ku, Tokyo Inside Graphics Communications Laboratories, Inc.

Claims (8)

[Claims] 1. For each frame constituting a moving image of the interlaced scanning system, a plurality of types of image information contained in pixels constituting the frame are respectively supplied in a horizontal scanning direction and a vertical direction based on a predetermined sampling frequency. Including a format conversion step of extracting in the scanning direction, for each frame extracted by the format conversion step, an intra-frame predicted image coded based on intra-frame information and a temporally forward frame reference image A forward predictive image encoded as, and a bidirectional predictive image encoded as a reference image temporally forward and backward frames, respectively, a moving image compression method for generating, Image discrimination for discriminating a predicted image to be encoded among the intra-frame predicted image, the forward predicted image and the bidirectional predicted image And a current image frame block partially forming the frame, the current image first field block including the first field and the current image frame block including the second field. A current image frame block, a current image first field block and a current image second
First extraction of image information representative of pixels forming the reference image for a field block to set a coarse search density and a wide search area, respectively.
And a similar block most similar to the current image frame block, the current image first field block and the current image second field block based on the search region set by the first search region setting step. A first motion that identifies and searches for a frame motion vector corresponding to the current image frame block, a first field motion vector corresponding to the current image first field block, and a second field motion vector corresponding to the current image second field block. A vector search step, and the first search in the vicinity of each of the similar blocks of the current image frame block, the current image first field block and the current image second field block specified by the first motion vector search step. Finer than the search area set by the area setting process. A second search area setting step of setting a narrow search area having a high search density, and the current image frame block and the current image first step based on the respective search areas set by the second search area setting step. The similar block that is most similar to the 1-field block and the 2nd field block of the current image is specified,
A second motion vector searching step of searching for a frame motion vector corresponding to the current image frame block, a first field motion vector corresponding to the current image first field block, and a second field motion vector corresponding to the current image second field block And the frame motion vector searched in the first motion vector search step, the first field motion vector,
A first motion vector determination step of determining an optimum motion vector from the second field motion vectors, the frame motion vector searched in the second motion vector search step, a first field motion vector,
A second motion vector determination step of determining an optimum motion vector from the second field motion vectors; a first motion vector determination step and a second motion vector according to the predicted image determined by the image determination step. And a selecting step of selecting one of the judging steps. 2. When the predicted image determined by the image determining step is a forward predictive image, the selecting step selects the second motion vector determining step, and the predicted image determined by the image determining step is For bidirectional predictive images,
The moving image compression method according to claim 1, wherein the selection step selects the first motion vector determination step. 3. A plurality of types of image information contained in pixels forming the field for each field forming a moving image of the interlaced scanning system are respectively scanned in a horizontal scanning direction and a vertical direction based on a predetermined sampling frequency. Including a format conversion step of extracting in the scanning direction, for each field extracted by the format conversion step, an intra-field predicted image coded based on intra-field information and a temporally forward field reference image A forward prediction image encoded as, and a bidirectional prediction image encoded as a reference image temporally forward and backward fields, respectively, a moving image compression method, The predicted image to be coded among the intra-field predicted image, the forward predicted image, and the bidirectional predicted image is An image discrimination step of discriminating, and a current image field block partially constituting the field is a first segment block of the current image composed of blocks in the upper half of the current image field block and a block in the lower half of the current image field block. A current image field block, a current image first segment block, and a current image second segment block, the image information representing the pixels forming the reference image is extracted, A first search area setting step of setting a wide search area having a coarse search density; and the current image field block and the current image first segment based on the search area set by the first search area setting step. Most similar to block and current image 2nd segment block Field motion vector similar block to identify the, corresponding to the current image field blocks,
A first motion vector searching step of searching for a first segment motion vector corresponding to the current image first segment block and a second segment motion vector corresponding to the current image second segment block, and the first motion vector searching step In the vicinity of the similar blocks of the identified current image field block, current image first segment block and current image second segment block, with a search density finer than the search region set by the first search region setting step. A second search area setting step of setting existing and narrow search areas respectively, and the current image field block, the current image first segment block, and the current image field block based on the respective search areas set by the second search area setting step. Similar block most similar to the second segment block of the current image A second motion vector for identifying and searching for a field motion vector corresponding to the current image field block, a first segment motion vector corresponding to the current image first segment block, and a second segment motion vector corresponding to the current image second segment block. A motion vector search step, and a first motion vector judgment that determines an optimum motion vector from the field motion vector, the first segment motion vector, and the second segment motion vector searched by the first motion vector search step. And a second motion vector determination step of determining an optimal motion vector from the field motion vector, the first segment motion vector, and the second segment motion vector searched by the second motion vector search step, Discriminated by the image discrimination process A selection step of selecting one of the first motion vector determination step and the second motion vector determination step according to the predicted image
A moving image compression method comprising: 4. If the predicted image determined by the image determining step is a forward predictive image, the selecting step selects the second motion vector determining step, and the predicted image determined by the image determining step is For bidirectional predictive images,
4. The moving image compression method according to claim 3, wherein the selecting step selects the first motion vector determining step. 5. For each frame forming a moving image of the interlaced scanning system, a plurality of types of image information contained in pixels forming the frame are respectively supplied in a horizontal scanning direction and a vertical direction based on a predetermined sampling frequency. A format conversion unit for extracting in the scanning direction is provided, and for each frame extracted by the format conversion unit, an intra-frame predicted image encoded based on the intra-frame information and a temporally forward frame are referred to. In a moving image compression apparatus, the forward predictive image encoded as an image and the bidirectional predictive image encoded as temporally forward and backward frames respectively as reference images are switched and generated. Image discrimination for discriminating a predicted image to be encoded among the intra-frame predicted image, the forward predicted image and the bidirectional predicted image Means, the frame comprises a first field and a second field, and a current image frame block partially forming the frame comprises a current image first field block comprising the first field and a current image first field block comprising the second field. A current image frame block, a current image first field block and a current image second
First extraction of image information representative of pixels forming the reference image for a field block to set a coarse search density and a wide search area, respectively.
And a similar block most similar to the current image frame block, the current image first field block and the current image second field block based on the search region set by the first search region setting unit. A first motion that identifies and searches for a frame motion vector corresponding to the current image frame block, a first field motion vector corresponding to the current image first field block, and a second field motion vector corresponding to the current image second field block. Vector search means, and the first search near the similar blocks of the current image frame block, current image first field block and current image second field block specified by the first motion vector search means. Finer than the search area set by the area setting means Second search area setting means for setting a narrow search area having a search density, and the current image frame block and the current image first based on the respective search areas set by the second search area setting means. The similar blocks that are most similar to the field block and the second field block of the current image are specified,
Second motion vector search means for searching a frame motion vector corresponding to the current image frame block, a first field motion vector corresponding to the current image first field block, and a second field motion vector corresponding to the current image second field block A frame motion vector searched by the first motion vector search means, a first field motion vector,
A first motion vector determination means for determining an optimum motion vector from the second field motion vectors, the frame motion vector searched by the second motion vector search means, a first field motion vector,
Second motion vector determination means for determining an optimal motion vector from the second field motion vectors, and first motion vector determination means and second motion vector according to the predicted image determined by the image determination means. A moving image compression apparatus, comprising: a determination unit that selects one of the determination units. 6. The predictive image determined by the image determining means is a forward predictive image, the selecting means selects the second motion vector determining means, and the predictive image determined by the image determining means is For bidirectional predictive images,
The moving picture compression apparatus according to claim 5, wherein the selection means selects the first motion vector determination means. 7. A plurality of types of image information contained in pixels forming the field for each field forming a moving image of the interlaced scanning method are respectively supplied in the horizontal scanning direction and the vertical direction based on a predetermined sampling frequency. A format conversion unit for extracting in the scanning direction is provided, and for each field extracted by the format conversion unit, an intra-field predicted image encoded based on the intra-field information and a temporally forward field are referred to. In a moving image compression apparatus, which generates by switching a forward prediction image encoded as an image and a bidirectional prediction image encoded by temporally forward and backward fields as reference images respectively, The predicted image to be encoded is determined from the intra-frame predicted image, the forward predicted image, and the bidirectional predicted image. Another image discriminating means, and a current image field block which partially constitutes the field is a current image first segment block consisting of an upper half block of the current image field block and a lower half block of the current image field block. A current image field block, a current image first segment block, and a current image second segment block, the image information representing the pixels forming the reference image is extracted, First search area setting means for setting a wide search area having a coarse search density, and the current image field block and the current image first segment based on the search area set by the first search area setting means Most similar to block and second segment of current image Field motion vector similar block to identify, corresponding to the current image field blocks,
A first motion vector searching means for searching a first segment motion vector corresponding to the current image first segment block and a second segment motion vector corresponding to the current image second segment block, and the first motion vector searching means In the vicinity of the similar blocks of the specified current image field block, current image first segment block and current image second segment block, with a search density finer than the search area set by the first search area setting means. Second search area setting means for setting a narrow search area, respectively, and the current image field block, the current image first segment block, and the current image field block based on the respective search areas set by the second search area setting means. The similar block most similar to the second segment block of the current image Searching for a field motion vector corresponding to the current image field block, a first segment motion vector corresponding to the current image first segment block, and a second segment motion vector corresponding to the current image second segment block. A second motion vector searching means, and a first motion for determining an optimum motion vector from the field motion vector, the first segment motion vector, and the second segment motion vector searched by the first motion vector searching means. Vector determining means, and second motion vector determining means for determining an optimum motion vector from the field motion vector, the first segment motion vector, and the second segment motion vector searched by the second motion vector searching means. And is discriminated by the image discrimination means. Selecting means for selecting either one of the first motion vector determining means and the second motion vector determining means according to the predicted image,
A moving image compression apparatus comprising: 8. If the predicted image determined by the image determination means is a forward predicted image, the selection means selects the second motion vector determination means, and the predicted image determined by the image determination means is For bidirectional predictive images,
8. The moving image compression apparatus according to claim 7, wherein the selection unit selects the first motion vector determination unit.