JP4507896B2 - Moving picture compression apparatus and moving picture compression processing method - Google Patents

Description

  The present invention relates to a moving image compression apparatus and a moving image compression processing method that perform moving image compression using motion compensation coding.

  In recent years, an encoding method using inter-frame correlation such as MPEG2 (ITU-T H.262) has been used as a moving image encoding method. In these methods, the image to be encoded is divided into encoded blocks that are small rectangular areas, a prediction block is obtained from the motion vector detected from the reference image for each encoding block, and the difference between the encoding block and the prediction block is compressed. A motion compensation coding method is used.

  FIG. 9 shows the relationship between MPEG2 video frames and references. Video frames are classified into three types: I picture, P picture, and B picture. The I picture is subjected to intra-frame compression called intra, but the P picture uses the past I picture or the past P picture as the reference screen, the B picture uses the past I or P picture as the forward reference screen, and the future I picture Alternatively, each motion vector is obtained using a P picture as a backward reference screen.

  As a typical motion vector search method, there is a block matching method. The block matching method is to extract a predicted block candidate having the same size as the encoded block from the motion vector search area, and to determine whether the extracted predicted block candidate is appropriate for use as a predicted block. This is a method of evaluating by calculating the error amount. All prediction block candidates in the motion vector search region are evaluated, and the one with the best evaluation result is adopted as the prediction block, and the difference in position coordinates between the adopted prediction block and the coding block is the motion vector. Become.

  The motion compensation coding method is a method for realizing highly efficient coding by greatly reducing the amount of information by taking the difference between the video data of the prediction block and the video data of the coding block obtained in this way. is there. Therefore, the image quality of the compressed video greatly depends on whether or not the motion vector search process can find a prediction block that matches the coding block well.

  Recently, moving picture compression using inter-frame correlation such as MPEG has been introduced in the field of video cameras. The video camera field is characterized by the fact that the entire screen often moves greatly, such as intense panning and tilting, in the video handled. However, when the movement of the entire screen exceeds the search range, it is extremely difficult to find an appropriate prediction block. For this reason, the image quality is unavoidably deteriorated, and simply increasing the search range as a countermeasure against it results in an increase in processing amount and an increase in power. Therefore, the method of setting the search area for the motion vector search is an issue for the video whose entire screen moves greatly, and improvement measures are being studied.

  As an example of a typical improvement measure, one described in Patent Document 1 is known. Now, a description will be given by taking as an example the case of forward reference where a P picture is an encoding screen and an I picture is a reference screen. FIG. 10 is a relationship diagram illustrating a method for setting a motion vector search range. First, the shift amount of the entire screen of the P picture and the I picture is obtained by some method, and is defined as an entire vector. A rectangle indicated by a broken line in FIG. 10 is a search area centered on a block having the same coordinates as the encoded macro block on the reference screen. If the entire vector is zero, this wavy line area becomes a motion vector search range. Now, when the entire vector is in the state shown in FIG. 10, the motion vector search range is shifted according to the entire vector, and a rectangular area indicated by hatching is set as the search area. With this motion vector search range shift function, a motion vector search can be performed in an appropriate search area even when the entire screen is shifted, such as panning and tilting. Further, the search range may exceed the range of the reference screen due to the shift of the motion vector search range. In that case, as shown in FIG. 11, the region where the reference screen exists is assumed as the actual search range in the search range assuming that the whole vector is zero, or the whole vector is shifted as shown in FIG. The area where the reference screen exists is adopted as the actual search range.

In Japanese Patent Application Laid-Open No. 2004-260260, a method for processing both camera shake correction and a motion vector search range setting method is described. However, in order to clarify the comparison with the present invention, only a method for setting a motion vector search range is described in a simplified manner. did. FIG. 10 is a simplified representation of FIG. 3 of Patent Document 1, and FIGS. 11 and 12 illustrate the contents described in Paragraph 0044 of Patent Document 1.
Japanese Patent Laid-Open No. 10-51787

  As described above, in a configuration in which the motion vector search range is not shifted, an extremely large search range is required for an image in which the entire screen is greatly shifted, such as panning and tilting, and the processing amount increases. There is a problem that electric power is required or processing time is increased.

  In a configuration such as Patent Document 1 that shifts the search range, an appropriate motion vector can be found as long as all pixels on the screen are shifted uniformly. However, in many cases, an image in which a large pan / tilt actually occurs is a shooting that tracks a moving object, and generally includes an image portion that moves differently from the movement of the entire screen. In these videos, even if the search range is shifted, it is difficult to find an appropriate motion vector because the search range is insufficient, and there is a problem that image quality degradation occurs.

  In addition, since there is no corresponding video portion in the reference screen for a video portion that has newly entered the encoded screen by pan / tilt, the possibility of finding an appropriate prediction block is extremely low. For this reason, the degree of information reduction is poor, and the motion vector also takes various values, so that the amount of motion vector coding is large, and as a result, the compression efficiency by the motion compensation method is poor. The region shown in FIGS. 11 and 12 corresponds to this. That is, in a video in which a large pan / tilt occurs, the motion compensation coding method is a process with poor compression efficiency that does not contribute to high-efficiency compression, resulting in a problem that image quality deteriorates.

  In view of the above problems, the present invention provides a moving image compression apparatus and a moving image compression processing method that reduce wasteful processing with poor compression efficiency, prevent a search range from being insufficient, and prevent an increase in processing amount. is there.

In order to solve this problem, the moving image compression apparatus according to claim 1 and the moving image compression processing method according to claim 3 of the present invention provide an area of an encoded screen that does not have a correspondence relationship based on the whole vector in the reference screen. Special region determination means or determination processing for determining a special region, and encoding method determination means or determination processing for determining whether to use a motion compensation method based on a motion vector or another method as an encoding method of a coding block And having the function of reducing the probability of adopting the motion compensation method as the coding method of the coding block belonging to the special region.

The moving image compression apparatus according to claim 2 and the moving image compression processing method according to claim 4 of the present invention are characterized in that, in the above invention, the special region is used only when the size of the entire vector is equal to or larger than a set reference value. Special area determination means for identifying or determination processing is provided.

  The moving image compression apparatus and the moving image compression processing method according to the present invention reduce only useless processing with poor compression efficiency and can expect an image quality improvement effect for a video whose entire screen is greatly shifted, such as pan and tilt. Since the encoding method that increases the processing amount and prevents erroneous processing judgment and is expected to be efficient is adopted, good video compression is achieved while preventing an increase in circuit scale and a decrease in processing speed. Can be realized.

According to the first aspect of the present invention, there is provided an entire vector detecting means for obtaining an entire vector which is a shift amount between the encoded screen and the reference screen as a whole, and an encoded screen having no correspondence relationship based on the entire vector in the reference screen. Special area determination means for determining the area as a special area, motion vector search means for searching for a motion vector of an encoded block in the encoding screen, and a prediction block and an encoding block based on a motion vector obtained by the motion vector search means A motion compensation evaluation means that evaluates the amount of information of the prediction error, which is the difference between the two, and obtains an evaluation value, and a motion compensation method using a motion vector is used as a coding method of a coding block based on the evaluation value, or another method is used. An encoding method determining means for determining whether or not the encoding method determining means is a motion compensation method as an encoding method for an encoding block belonging to a special region. Is intended to reduce the probability of adopting, thereby having the effect of employing a motion compensation scheme only to the coding blocks reliably efficiency expected good in regions compression efficiency by the motion compensation is poor.

The invention according to claim 2 has the function of identifying as a special region only when the size of the entire vector is equal to or larger than a set reference value, in addition to the configuration according to claim 1. According to the first aspect of the present invention, there is a possibility that a side effect of increasing the processing amount as compared with the degree of image quality improvement is caused only when the whole vector is small. It has an action to prevent.

The moving image compression processing method according to the third and fourth aspects of the invention is a processing method corresponding to the moving image compression device according to the first and second aspects, respectively, and has the same operation as the corresponding claim. Have.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

(Embodiment 1)
FIG. 1 is a block diagram showing a configuration of a moving image compression apparatus according to Embodiment 1 of the present invention. In FIG. 1, an encoding screen memory 1 and a reference screen memory 2 supply video data to an entire vector detection means 3 and a motion vector search means 5, and an output of the entire vector detection means 3 is sent to a special area determination means 4 and a motion vector. The output of the search means 5 is connected to the motion compensation evaluation means 6 and the NoMC evaluation means 7, respectively. The encoded screen memory 1 supplies video data to the intra evaluation unit 8, and the evaluation results of the motion compensation evaluation unit 6, NoMC evaluation unit 7, and intra evaluation unit 8 are connected to the encoding method determination unit 9. The special area determination unit 4 provides search range information to the motion vector search unit 5 and priority information to the encoding method determination unit 9.

  Hereinafter, the operation of the present embodiment will be described. The operation in the first embodiment is an operation in the case of encoding the P picture shown in FIG. 9, that is, a moving image compression operation of the MPEG2 standard using forward reference. It is assumed that an encoded P picture is already stored in the reference screen memory 2 in the I picture or P picture corresponding to the past three frames of the encoded P picture. In this embodiment, the operation differs depending on whether the entire screen movement such as pan / tilt is large and small or no movement. Hereinafter, it demonstrates in order.

  First, a case where the entire screen movement is small will be described.

  First, the whole vector detection means 3 detects the movement state of the entire screen of the encoding screen and the reference screen and outputs it as a whole vector. As a method for detecting the whole vector, a method for obtaining the total movement by searching the moving state of several parts of the screen, generating a reduced screen for each of the encoded screen and the reference screen, and searching for a shifted state of the reduced screen. There is a method of obtaining a camera shake vector by a camera shake correction gyro instead of video data, and synthesizing it. The whole vector detection method is not the main point of the present invention, and any method may be used. FIG. 2 is an explanatory diagram showing the definition of the entire vector in the present embodiment. FIG. 2 illustrates a state where the lower right rectangular area of the encoding screen and the upper left rectangular area of the reference screen are in good agreement. In this case, the entire video is moving in the lower right direction on the screen in time, and it can be estimated that the camera has moved to the upper left during shooting. The definition of the whole vector used in the present embodiment is shown on the encoding screen in FIG. 2, and corresponds to the movement vector of the camera.

  The special area determination means 4 receives the whole vector from the whole vector detection means 3, and classifies and identifies the encoded screen into two areas, a special area and a normal area. FIG. 3 is an explanatory diagram showing the relationship between the special area and the normal area. The special area is considered to be a video portion that has newly entered after the time of the reference screen on the encoded screen. All areas other than the special area are treated as normal areas. Further, when the entire vector is smaller than a preset reference value, the special area determination unit 4 determines that the special area does not exist and the entire encoded screen is the normal area. Since the operation when the entire movement is small, that is, when the entire vector is small, is considered, the special area determination unit 4 identifies that the entire encoded screen is the normal area.

  When the above-described classification and classification of the region is confirmed, the motion vector search means 5 starts a motion vector search operation. The motion vector search operation divides the encoded screen into encoded macroblocks that are rectangular areas of 16 pixels in the vertical direction and 16 pixels in the horizontal direction, sets the motion vector search area on the reference screen for each encoded macroblock, In this search area, an optimum prediction block that best matches the encoded macroblock is searched for by block matching. Since the entire vector is now small, the special area determination means 4 instructs the motion vector search means 5 using the standard search range as search range information. The standard method for determining the position and size of the search range is not the gist of the present invention, and any method can be introduced. In the first embodiment, it is assumed that the conventional search range shown in FIG. 10 is adopted.

  The motion vector search means 5 transmits the prediction block data and the encoded macroblock data determined to be optimal as the search results to the motion compensation evaluation means 6, and the motion compensation evaluation means 6 transmits the prediction block data and the encoded macroblock data. An evaluation operation is performed to obtain a data difference and predict an information amount when the difference is used as encoded data. The evaluation result is transmitted to the encoding method determining means 9 as a motion compensation evaluation value. Also, the motion vector search means 5 cuts out a prediction block when it is assumed that the motion vector is zero from the reference screen, and transmits the data of the prediction block and the data of the encoded macro block to the NoMC evaluation means 7. The NoMC evaluation means 7 predicts the amount of information when the difference between the prediction block data and the encoded macroblock data is used as the encoded data, and transmits it to the encoding method determination means 9 as the NoMC evaluation value. Further, the intra evaluation unit 8 reads out the encoded macroblock from the encoding screen memory 1, predicts the amount of information when it is used as encoded data, and transmits it to the encoding method determination unit 9 as an intra evaluation value.

  The encoding method determining means 9 compares and judges the transmitted motion compensation evaluation value, NoMC evaluation value, and intra evaluation value, and adopts any method of motion compensation encoding, NoMC encoding, or intra encoding. Or for each macroblock. Since the entire vector is now small, the special area determination unit 4 instructs the encoding method determination unit 9 that the priorities are all equal as priority information. In response to the instruction, the encoding method determining means 9 decides to adopt a method that simply compares the motion compensation evaluation value, the NoMC evaluation value, and the intra evaluation value and shows the best numerical value. Hereinafter, the encoded data is orthogonally transformed according to the selected method, and high-efficiency encoding is performed.

  Next, a case where the movement of the entire screen is large and the entire vector is large will be described.

  When the whole vector is large, the special area determination unit 4 first determines that the special area shown in FIG. 3 exists, and classifies and identifies the encoded screen into two areas, a special area and a normal area. This is the operation corresponding to the fourth aspect. When the entire vector is large, it is very unlikely that an encoded macroblock belonging to the special area in FIG. 3 has a suitable video as a prediction block in the reference screen. It is very likely that good motion compensation cannot be realized. Therefore, the special area determination means 4 gives an instruction to omit the motion vector search to the motion vector search means 5 with respect to the encoded macroblock belonging to the special area as search range information, and does not execute the motion vector search processing. This is the operation corresponding to the first aspect. For example, when the size of the entire vector is as shown in FIG. 3, about 30% of the processing amount of the motion vector search operation is reduced. The special area determination unit 4 gives priority information to the encoding method determining unit 9 so as not to adopt the motion compensation method, and the encoding method determining unit 9 selects either NoMC encoding or intra encoding. . In general, when intra coding is selected, the amount of code increases, but the video is blurred in a special area when the entire vector is large, and since it is an area at the edge of the screen, it is unlikely to include a point of interest. Even when a method is selected, compression with a reduced code amount can be realized. By adopting the configuration of claim 1 in this way, it is possible to greatly reduce the amount of motion vector search processing while minimizing the influence of coding efficiency.

  On the other hand, since the normal region encoded macroblock that does not belong to the special region in FIG. 3 is very likely to have an appropriate region as a predicted image in the reference screen, the motion vector search process is effectively and efficiently performed. It can be expected that it contributes to a correct encoding. However, as described above, when a large pan or tilt occurs, tracking shooting is often performed with a moving object as a point of interest. When the attention point moves greatly different from the entire vector, it is highly possible that the attention point is included in the normal region. In addition, the point of interest is that if an appropriate prediction block can be found, the amount of codes can be reduced without losing the image quality, which is also a part that is highly effective for high-efficiency coding. Therefore, the special area determination means 4 gives search range information to the motion vector search means 5 for the encoded macroblock in the normal area, and expands the motion vector search area. This is an operation corresponding to claim 2. FIG. 4 is an explanatory diagram of the search area expansion. The search range before enlargement in FIG. 4 is the standard search range described above when the motion vector is small, and is expanded to an area twice as large as the entire vector. In this way, by expanding the search range of the encoded macroblock in the normal area, the attention point that moves differently from the whole vector can be detected well as a prediction block, and the image quality can be improved while suppressing the code amount. It can be done. FIG. 4 shows a case where the search area is approximately doubled. In the first embodiment, the processing amount of the special area is reduced by, for example, about 30% in claim 1, so that the entire motion vector search is performed. The amount of processing is within an increase of about 40%. Since the special area determination means 4 can obtain the processing amount reduction ratio by the special area before the start of the motion vector search operation, if it is determined that the reduction is 30%, the expansion of the search area can be set to 1.43 times. The total amount of motion vector search processing can be kept constant.

  In the first embodiment, the case where the special region determination unit 4 omits the motion vector search operation in the encoded macroblock belonging to the special region in FIG. 3 has been described. The motion vector search means 5 may be instructed to reduce the motion vector search area for the encoded macroblocks belonging to. This is also a technique corresponding to claim 1. FIG. 5 is an explanatory diagram of the search area reduction. When the search area is shifted by the entire vector, the encoded macroblock in the special area always protrudes outside the range of the reference screen. In FIG. 5, a rectangle having the same coordinates as that of the encoded macroblock is shifted by the whole vector, then vertically shifted so as to enter the inside of the reference screen, and a small reduced search range is provided around the rectangle. Since the reduced search range is sufficiently smaller than the standard search range, it is possible to realize a significant reduction in processing amount. In some cases, for example, when the video to be compressed is an animation or the like and a background image in which the special area is patterned, even a coded block belonging to the special area can easily obtain a very good prediction block. Therefore, a good prediction block can be obtained efficiently by providing a small search range as shown in FIG.

  On the other hand, as described above, when the video to be compressed is a natural image, an encoded macroblock belonging to the special area cannot obtain a good prediction block. However, it is extremely difficult to accurately estimate the code amount in each case by the motion compensation evaluation means 6, NoMC evaluation means 7, and intra evaluation means 8, and in many cases, it is inappropriate even in a special region. Motion compensation is employed, which causes an increase in code amount and deterioration in image quality. Therefore, when the motion vector search is also executed for the coding macroblock in the special region, the special region determination unit 4 lowers the priority of the motion compensation evaluation value in the special region to the coding method determination unit 9 based on the priority information. Let them be weighted as follows. This corresponds to the third aspect. Thereby, for example, motion compensation is adopted when a very good prediction block such as an animation is obtained, and another method is surely adopted when a good prediction block such as a natural image is not obtained.

  As described above, in the first embodiment, the motion vector search in the special area, which is an inefficient process, is greatly reduced by the configuration of claim 1, and the search range is expanded by the configuration of claim 2 and is effective. Only the search area is effectively enlarged, the structure of claim 3 prevents the inappropriate determination of the encoding method for the special area, and the structure of claim 4 makes the effects of claims 1, 2, and 3 remarkable. With certainty adapted to a certain case, it is possible to realize a good image quality while minimizing an increase in the processing amount even for a video with a large movement of the entire screen such as pan and tilt.

(Embodiment 2)
FIG. 6 is a block diagram showing the configuration of the moving picture compression apparatus according to Embodiment 2 of the present invention. In FIG. 6, the same components as those of the first embodiment shown in FIG. The encoded screen memory 1 and the reference screen memory 10 supply the video data to the whole vector detecting means 11 and the motion vector searching means 13, and the output of the whole vector detecting means 11 is sent to the special area judging means 12 and the motion vector searching means 13. The outputs are respectively connected to the motion compensation evaluation means 14. The outputs of the motion vector search means 5 and the motion vector search means 13 are connected to the bidirectional motion compensation evaluation means 15, and the motion compensation evaluation means 6, motion compensation evaluation means 14, bidirectional motion compensation evaluation means 15, intra evaluation means 8 The evaluation result is connected to the encoding method determination means 16. The special area determination unit 12 provides motion information to the motion vector search unit 13 and the encoding method determination unit 16, respectively.

  Hereinafter, the operation of the second embodiment will be described. The operation of the second embodiment is an operation in the case of encoding the B picture shown in FIG. 9, that is, an MPEG2 standard moving image compression operation using a bidirectional reference of a forward reference and a backward reference. The encoded screen picture 1 in the encoded screen memory 1, the past P picture of the encoded B picture as the reference picture in the forward direction in the reference picture memory 2, and the future as the reference picture in the reverse direction in the reference picture memory 10 It is assumed that each of the P pictures has already been stored.

  In the second embodiment, the whole vector detection means 3, the special region determination means 4, the motion vector search means 5, and the motion compensation evaluation means 6 use the video data stored in the coding screen memory 1 and the reference screen memory 2 to obtain the whole vector. Executing the forward-referenced motion vector search operation while applying is exactly the same as in the first embodiment. In addition, the entire vector detection unit 11, the special region determination unit 12, the motion vector search unit 13, and the motion compensation evaluation unit 14 apply the entire vector using the video data stored in the encoded screen memory 1 and the reference screen memory 10. A backward-referenced motion vector search operation is executed, which is also different from the reference screen memory used, and the operation content is exactly the same as that in the forward direction.

  The bidirectional motion compensation evaluation means 15 generates a bidirectional prediction block by combining the forward prediction block obtained from the motion vector search means 5 and the backward prediction block obtained from the motion vector search means 13 and combines them. The amount of information when the difference between the bidirectional prediction block and the encoded macroblock is used as encoded data is predicted, and a bidirectional motion compensation evaluation value is output.

  The encoding method determining means 16 is transmitted from the forward motion compensation evaluation value transmitted from the motion compensation evaluation means 6, the backward motion compensation evaluation value transmitted from the motion compensation evaluation means 14, and transmitted from the bidirectional motion compensation evaluation means 15. The bi-directional motion compensation evaluation value and the intra-evaluation value transmitted from the intra-evaluation means 8 are compared and determined to determine whether forward motion compensation coding, reverse motion compensation coding, bidirectional motion compensation coding, It is determined for each encoded macro block whether intra encoding is used. Hereinafter, the encoded data is orthogonally transformed according to the selected method, and high-efficiency encoding is performed.

  Here, the entire vector in the forward direction and the reverse direction will be described. FIG. 7 is an explanatory diagram of the entire vector in the forward direction and the reverse direction. In FIG. 7, the relationship between the forward reference screen, the encoding screen, and the entire forward vector is exactly the same as the relationship between the reference screen, the encoding screen, and the entire vector already described in FIG. Regarding the reverse direction, FIG. 7 illustrates a state in which the upper left rectangular area of the encoded screen and the lower right rectangular area of the reverse direction reference screen are in good agreement. In this case, it can be estimated that the entire video has moved in the lower right direction in the screen from the encoding screen toward the backward reference screen, and that the camera has moved to the upper left during shooting. The definition of the entire backward vector used in the present embodiment corresponds to the backward direction of the camera movement vector, as shown on the encoding screen in FIG.

  The special area determination means 4 and 12 independently set the motion vector search range according to claim 1 and claim 2 from the forward whole vector and backward whole vector, and the motion compensation evaluation value according to claim 3, respectively. Specify the priority. According to the fourth aspect of the present invention, the operation switching based on the magnitude relationship with the reference value of the entire vector is also performed independently in the forward direction and the reverse direction. Therefore, when both the forward direction overall vector and the backward direction overall vector are small, the encoding method determining means 16 simply calculates the forward direction motion compensation evaluation value, the backward direction motion compensation evaluation value, the bidirectional motion compensation evaluation value, and the intra evaluation value. It will be decided to adopt the method that shows the best numerical value by comparison. Also, if either one of the forward global vector and the backward global vector is larger than the reference value and there is a special region, for example, assuming that the forward global vector is large, the encoded macroblock belonging to the forward special vector is The priority of the forward motion compensation evaluation value is set low, and as a result, the backward motion compensation method or the intra method is adopted.

  Finally, the operation in the case where both the forward general vector and the backward general vector are large and both have a special region will be described. FIG. 8 comprehensively shows the forward and backward area divisions of the special area determination means 4 and 12. In FIG. 8, area A belongs to the normal area in both the forward and reverse directions, and the encoded macroblocks in this area are expanded in both the motion vector search means 5 and 13, and the encoding method determining means 16 This is an area where the evaluation values are evaluated equally and the most efficient encoding is realized. In the region B, either the forward direction or the reverse direction is a special region, and the encoded macroblock has a motion vector search region reduced or omitted on the side corresponding to one of the special regions of the motion vector search means 5 and 13. On the other hand, since the presence of an appropriate prediction block can be expected, the expansion of the search area is instructed. Furthermore, since the priority of the motion compensation evaluation value corresponding to the special region is lowered in the encoding method determining means 16, the motion compensation method or intra method corresponding to the normal region is preferentially adopted. In the area C, both the forward and reverse directions are special areas, and the motion vector search means 5 and 13 reduce or omit the motion vector search area for the encoded macroblocks in this area. Furthermore, since the priority of the motion compensation evaluation value is all lowered in the encoding method determining means 16, the intra method is preferentially adopted.

  As described above, in the second embodiment, the present invention can be applied to a bi-directional reference method such as the B picture of the MPEG2 standard, and the motion vector search in the special area, which is an inefficient process, is performed by the configuration of claim 1. The search range is greatly reduced, the search range is expanded by the configuration of claim 2, and the search region is effectively expanded only by the effective region. The configuration of claim 3 prevents inappropriate determination of the encoding method for the special region. In addition, the configuration of claim 4 provides certainty for adapting to the case where the effects of claims 1, 2, and 3 are remarkable, and the processing amount is increased even for images such as pan and tilt where the entire screen moves rapidly. It is possible to achieve good image quality while minimizing the image quality.

  In the first embodiment and the second embodiment, the first, second, third, and fourth embodiments have been described as being applied in combination. However, the first, second, and third embodiments can be individually adapted to achieve individual effects. It is obtained. Claims 5, 6, 7, and 8 are processing methods corresponding to claims 1, 2, 3, and 4, respectively, and some or all of the processing functions described above can be realized using computer resources. Needless to say, moving image compression software having the same effect can be realized by adapting these.

  The moving image compression apparatus and the moving image compression processing method according to the present invention reduce the processing with poor compression efficiency and increase the processing amount only in a video region where an image quality improvement effect can be expected for a video whose entire screen is largely shifted. In addition, motion compensation coding such as MPEG can be realized by preventing an erroneous processing judgment and realizing an excellent video compression while preventing an increase in circuit scale and a reduction in processing speed. It can be applied to any device or software that uses video to compress moving images.

1 is a block diagram showing the configuration of a moving image compression apparatus according to Embodiment 1 of the present invention. Explanatory diagram of the entire vector in the moving image compression apparatus Explanatory drawing of special area in the moving image compression apparatus Explanatory drawing of search area expansion in the moving image compression apparatus Explanatory drawing of search area reduction in the moving image compression apparatus A block diagram showing a configuration of a moving image compression apparatus according to a second embodiment of the present invention. Explanatory diagram of the entire vector in the moving image compression apparatus Explanatory drawing of special area in the moving image compression apparatus Conceptual diagram for explaining the reference relationship of MPEG2 Conceptual diagram showing the relationship between the entire vector and the motion vector search range in a conventional video compression apparatus Conceptual diagram showing the relationship between the entire vector and the motion vector search range in a conventional video compression apparatus Conceptual diagram showing the relationship between the entire vector and the motion vector search range in a conventional video compression apparatus

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Encoding screen memory 2 Reference screen memory 3 Whole vector detection means 4 Special area determination means 5 Motion vector search means 6 Motion compensation evaluation means 7 NoMC evaluation means 8 Intra evaluation means 9 Encoding method determination means 10 Reference screen memory 11 Whole vector Detection means 12 Special region determination means 13 Motion vector search means 14 Motion compensation evaluation means 15 Bidirectional motion compensation evaluation means 16 Coding method determination means

Claims (4)

A whole vector detecting means for obtaining a whole vector that is a deviation amount of the whole screen of the encoding screen and the reference screen;
Special region determination means for determining a region of the encoded screen that does not have a correspondence relationship based on the whole vector as a special region;
Motion vector search means for searching for a motion vector of a coding block in the coding screen;
Motion compensation evaluation means for evaluating an information amount of a prediction error that is a difference between a prediction block and a coding block by a motion vector obtained by the motion vector search means and obtaining an evaluation value;
Coding method determining means for determining whether to use a motion compensation method based on the motion vector or another method as a coding method of a coding block based on the evaluation value;
The encoding scheme determining means reduces the probability of adopting the motion compensation scheme as the encoding scheme of the encoding block belonging to the special region;
Video compression device. The special area determination means identifies the special area only when the size of the whole vector is equal to or larger than a set reference value .
The moving image compression apparatus according to claim 1 . Obtaining an overall vector, which is a deviation amount of the entire screen of the encoding screen and the reference screen;
Identifying the area of the encoded screen that does not have a reference screen with the correspondence based on the whole vector as a special area;
A motion vector search step of searching for a motion vector of a coding block in the coding screen;
A step of evaluating an information amount of a prediction error which is a difference between a prediction block and a coding block by a motion vector obtained in the motion vector search step and obtaining an evaluation value;
A coding method determining step for determining whether to use a motion compensation method based on the motion vector or another method as a coding method of a coding block based on the evaluation value;
A control step for reducing the probability of adopting the motion compensation method as a coding method of a coding block belonging to the special region,
A moving image compression processing method. Added control to determine that a special area exists only when the size of the entire vector is greater than or equal to a set reference value ,
The moving image compression processing method according to claim 3 .

Images