JP2018117239A - Video data compression apparatus and video data compression method - Google Patents

Abstract

A moving image data compression apparatus and moving image data compression method capable of improving the compression rate of a moving image including transparency information are provided. A moving image compression apparatus for compressing moving image data configured using frame image data in which transparency information is set at least partially includes a specific frame image data and a front of the specific frame image data. A motion vector detection unit 17 that detects a motion vector between other frame image data existing in the direction and / or backward direction, and a code generation unit 13 that compresses moving image data using the detected motion vector; The motion vector detection unit 17 uses the information on the transparency of the specific frame image data and / or the other frame image data, and moves between the specific frame image data and the other frame image data. Detect vectors. [Selection] Figure 1

Description

  The present invention relates to a moving image data compression apparatus and moving image data compression method for compressing frame image data of moving image data in which transparency information is set.

  Conventionally, inter-frame prediction is known as a technique used when compressing moving image data (see, for example, Patent Document 1). This is based on the encoding target frame (for example, the latest frame among frames arranged in chronological order) and the reference frame at a different time from the encoding target frame (for example, the past frame one before the latest frame). This is a technique for reducing the data amount of moving image data by encoding a difference image with a generated predicted image.

JP 2010-200377 A

  Here, as processing when encoding the inter-frame difference in inter-frame prediction, the image information included in the encoding target frame and the image information included in the reference frame are digitized, and the image is processed by processing based on a predetermined calculation. It is conceivable to obtain a motion vector as the direction and size of information motion and encode the obtained motion vector.

  On the other hand, in recent years, image information includes color information (for example, R, G, and B information forming an RGB color space, and Y and Y converted from luminance information (Y) and color difference information (U, V)). In addition to information on U, V, etc., transparency information that defines transparency (transparency of other superimposed images when a plurality of images are superimposed) may be included. If the image information has transparency information in addition to the color information, the characteristics of the image information may differ from the case of only the color information, so the same processing as the calculation and processing for the color information is performed. That is not always appropriate. That is, when the encoding target frame and the reference frame include a transparent part or a part with high transparency, the color information of the encoding target frame or the reference frame is more suitable for determining the motion vector than the opaque image information. In many cases, the degree of importance in computation and processing is not high.

  However, in the above-mentioned Patent Document 1, no consideration is given to the case where the image information has transparency, or the case where the image information has different degrees of transparency. For this reason, in Patent Document 1, a motion vector is obtained by the same calculation and processing for a transparent image as for an opaque image. Therefore, in Patent Document 1, even if the importance of color information is not high for moving image data having a transparent portion, a motion vector is obtained by performing calculation and processing centering on the color information. . This unnecessarily increases the amount of computation and unnecessarily increases the amount of code, which has the problem of inhibiting improvement in the compression rate of moving images.

  The present invention has been made in view of such problems, and an object of the present invention is to provide a moving image data compression device and a moving image data compression method capable of improving the compression rate of moving images including transparency information. .

  In order to solve such a problem, the invention described in claim 1 is a moving image data compression device that compresses moving image data configured using frame image data in which transparency information is set at least in part. A motion vector detecting means for detecting a motion vector between a specific frame image and other frame image data existing in the forward direction and / or backward direction of the specific frame image, and the detected motion Compression means for compressing the frame image data using a vector, and the motion vector detection means uses the transparency information of the specific frame image data and / or the other frame image data. The motion vector between the specific frame image data and the other frame image data is detected.

  According to a second aspect of the present invention, in addition to the configuration according to the first aspect, the motion vector detecting means may include the transparency information of the specific frame image data and / or the other frame image data. The motion vector is detected by correcting the contribution of image quality in determining the motion vector depending on the value of the transparency level.

  According to a third aspect of the present invention, in addition to the configuration of the second aspect, the motion vector detecting means defines the transparency information of the specific frame image data when detecting the motion vector. As the transparency and / or the transparency specified by the transparency information of the other frame image data is higher, the color information of the specific frame image data and / or the color information of the other frame image data. The influence on the determination of the motion vector is made small.

  According to a fourth aspect of the present invention, in addition to the configuration according to any one of the first to third aspects, the motion vector detecting means includes the color information of the specific frame image data in the color information of the specific frame image data. Detecting the motion vector using a value obtained by multiplying the transparency information and / or a value obtained by multiplying the color information of the other frame image data by the transparency information of the other frame image data. Features.

  According to a fifth aspect of the present invention, in addition to the configuration of the fourth aspect, the motion vector detecting means multiplies the color information of the specific frame image data by the transparency information of the specific frame image data. Detecting the motion vector using an absolute value error or a square error of a value and / or a value obtained by multiplying the color information of the other frame image data by the transparency information of the other frame image data. Features.

  According to a sixth aspect of the present invention, in addition to the configuration according to any one of the first to fifth aspects, the motion vector detecting means includes transparency information of the specific frame image data and / or the other The motion vector is detected using transparency information of frame image data.

  The invention according to claim 7 is a moving image data compression method for compressing moving image data configured using frame image data in which transparency information is set at least in part, and the specific frame image data A motion vector detection procedure in which a motion vector is detected between the frame image data and other frame image data existing in the forward direction and / or backward direction of the specific frame image data, and using the detected motion vector A compression procedure for compressing the frame image data, and in the motion vector detection procedure, using the transparency information of the specific frame image data and / or the other frame image data, The motion vector between the specific frame image data and the other frame image data is detected.

  According to the first and seventh aspects of the invention, the specific frame image data and / or the other frame image data using the transparency information of the specific frame image data and / or other frame image data. By detecting the motion vector between the frame image data, the motion vector is detected depending on the set transparency information in the frame image of the moving image data in which the transparency information is set, and the frame image data is compressed. Can do. Since the motion vector can be detected depending on the transparency information, color information of low importance in detecting the transparent portion in the detection and encoding of the motion vector of the frame image data having the transparent portion is the motion vector. It is possible to reduce the degree of reflection in the detection and encoding. Thereby, the compression rate of the moving image containing transparency information can be improved.

  According to the second aspect of the invention, depending on the value of the transparency level, the motion vector is detected by correcting the contribution of the image quality in the determination of the motion vector. Can be reflected in the detection and encoding of motion vectors, and the compression rate of a moving image including transparency information can be improved.

  According to the third aspect of the present invention, the higher the transparency defined by the transparency information of the frame image data, the smaller the influence of the color information of the frame image data on the magnitude of the motion vector, As the transparency of the frame image data becomes higher and the importance of the color information becomes lower when detecting the motion vector, the degree to which the color information is reflected in the detection and encoding of the motion vector can be reduced. Thereby, in the compression of a moving image including transparency information, it is possible to improve the compression rate while reducing less important information.

  According to the invention described in claim 4, since the motion vector can be detected by correcting the value obtained based on the color information of the frame image data based on the transparency information, the transparency information is included. In moving image compression, the degree of transparency can be reflected in specific calculation processing in motion vector detection and encoding. Accordingly, it is possible to realize in a specific calculation process that the compression rate is improved while reducing the data with low importance based on the degree of transparency.

  According to the invention described in claim 5, by detecting a motion vector based on an absolute value error or a square error of a value obtained by multiplying color information by transparency information, accuracy can be determined based on a value reflecting transparency information. It is possible to detect an appropriate motion vector by performing a high calculation process.

  According to the invention described in claim 6, by detecting a motion vector using transparency information, high-precision arithmetic processing is performed based on a value reflecting the transparency information, and an appropriate motion vector is detected. can do.

It is a functional block diagram explaining the whole structure in the moving image compression apparatus which concerns on this embodiment. It is a functional block diagram which shows the detail of the prediction process part in a moving image compression apparatus same as the above. It is a figure which shows typically the principle of the calculation of the motion vector in a moving image compression apparatus same as the above. It is a figure which shows typically the correlation of the information of color information and transparency information in the calculation in the motion vector detection part in a moving image compression apparatus same as the above. It is a flowchart which shows an example of the process sequence in a moving image compression apparatus same as the above.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Although the present embodiment will be described based on an example applied to an MPEG encoding method, the present invention may be applied to any image encoding method other than MPEG.

[Movie data used in this embodiment]
The moving image data used in this embodiment is moving image data having transparency (α value) information. Specifically, for example, in addition to “R” information, “G” information, and “B” information in the RGB color space, as image information of each pixel of a frame as “frame image data” that constitutes moving image data. When the “A” information that defines the transparency (α value) is set with a value from 0 to 255 (however, in the “A” information, 0 is the maximum transparency (transparency 100%), 255 is the minimum transparency (transparency) 0%)) corresponds to this.

  In this embodiment, the moving image compression apparatus 10 (see FIG. 1) compresses a frame including this transparency information. Note that in the moving picture compression apparatus 10 (see FIG. 1), all frames including transparency (α value) information may be compression targets, or only a part may be compression targets. In addition, the moving image data to be compressed in the moving image compression apparatus 10 may have transparency (α value) information for all frames, or at least some of the frames may have transparency (α value). You may have information. Further, in the moving picture compression apparatus 10 (see FIG. 1), compression (described later) using transparency (α value) information and a compression method other than compression using transparency (α value) information are used in combination. The video data may be compressed.

[Basic configuration of video compression device]
FIG. 1 shows a functional block diagram of the moving picture compression apparatus 10 of this embodiment, and FIG. 2 shows a functional block diagram showing a prediction processing unit of the moving picture compression apparatus of this embodiment. The moving image compression apparatus 10 as the “moving image data compression apparatus” of this embodiment has a function as an encoder, and compresses and encodes moving image data having transparency information. Specifically, for example, in this video compression device, a plurality of separate videos are superimposed, and when they are superimposed, alpha blending is displayed according to the transparency, and a part of the video displayed on the rear side from that part is displayed. Is used for compression encoding of moving images of gaming machines, which are displayed on the display in a state where is transmitted. However, it may be used for compression encoding of any moving image having transparency information that is displayed in a superimposed manner and is used for other than gaming machines.

  In this moving image compression apparatus 10, moving image data is processed in units of a predetermined rectangular area, that is, in units of a matrix composed of a predetermined number of pixels (details will be described later).

  In the frame image data in which one image and the other image are superimposed, when compressing the image data according to the transparency, even if the frame image data of one image is compressed according to the transparency of the one image, The frame image data of the other image may be compressed, or the frame image data of both images may be compressed.

  This embodiment will be described using an example of an apparatus that compresses frame image data of one image in accordance with the transparency of one image.

  The moving picture compression apparatus 10 of this embodiment includes a prediction processing unit 11, a DCT unit 12, a code generation unit 13, and a frame memory 14 as a configuration for performing compression coding as an encoder. The moving image compression apparatus 10 also includes an inverse quantization unit 15 and an inverse DCT unit 16 as a configuration for performing a decoding process of compressed data as a decoder.

  The prediction processing unit 11 performs a process for performing inter-frame prediction using a plurality of frames as “moving image data”. Specifically, a predicted image (a moving image frame generated by performing a predetermined prediction process on a past moving image frame) is generated and output from moving image data as an input image. The predicted image is supplied to the DCT unit 12 as a prediction error generated by subtracting the input image, and is added to the DCT data generated by the inverse DCT unit 16 to generate a reference image, which is stored in the frame memory 14. Remembered.

  The DCT unit 12 performs DCT (discrete cosine transform) on each image data. Specifically, frequency conversion such as discrete cosine transformation is performed on each image data in units of blocks, thereby decomposing the frequency components into coefficientized DCT data. The DCT unit 12 of this embodiment generates DCT data by performing frequency conversion of the prediction error formed by the difference between the input image and the prediction image as a result of the inter-frame prediction processing in the prediction processing unit 11. To do. Note that the DCT unit 12 of this embodiment is an example of a configuration, and of course, wavelet transform or Hadamard transform is used instead of DCT (discrete cosine transform), DPCM (Differential Pulse Code Modulation: differential pulse) It is also possible to use (code modulation).

  The code generation unit 13 generates and encodes quantized data by quantizing the image data processed by a predetermined encoding method based on the quantization coefficient. Here, the “predetermined encoding scheme” refers to a transform encoding scheme based on frequency conversion such as discrete cosine transform in the DCT unit 12 which is the immediately preceding configuration.

  As shown in FIG. 1, the code generation unit 13 uses a quantization unit 19 that quantizes DCT data using quantization coefficients and the like to generate quantization data, and entropy coding such as Huffman code and arithmetic code. And a variable length encoding unit 20 for further generating the encoded data by further compressing the quantized data.

  The frame memory 14 is various storage media such as a RAM and a cache, and stores various moving image data. Here, moving image data generated as a result of inter-frame prediction processing in the prediction processing unit 11 is stored. The moving image data generated as a result of the inverse DCT process in the inverse DCT unit 16 is also stored.

  The inverse quantization unit 15 decodes the DCT data by inversely quantizing the input quantized data using a predetermined quantization coefficient. In this embodiment, the quantized data input to the inverse quantization unit 15 is output from the quantization unit 19 (see FIG. 1). The inverse DCT unit 16 performs inverse DCT processing on the DCT data output from the inverse quantization unit 15. The moving image data output from the inverse quantization unit is combined with the predicted image (described later) output from the prediction processing unit 11 to form a reference image (described later) as a past moving image frame, and is stored in the frame memory 14. The

  The prediction processing unit 11 includes a motion vector detection unit 17 as a “motion vector detection unit” and a motion compensation unit 18.

  The motion vector detection unit 17 detects a motion vector between a plurality of moving image frames. Specifically, the motion vector detection unit 17 detects a motion vector between specific frame image data and other frame image data existing in the forward direction and / or backward direction of the specific frame image data. To do. In this embodiment, a specific time, for example, the current frame image and the previous frame image are each divided into a plurality of substantially rectangular blocks, and a block serving as a reference position of the frame image at a specific time The image information (color information and transparency information) is searched from which block of the other frame image, and a motion vector as a movement direction and a movement amount of these blocks is detected.

  The motion compensation unit 18 performs a motion compensation process for moving a motion image by a motion vector from a reference image (past moving image frame).

  FIG. 2 is a functional block diagram illustrating details of the prediction processing unit 11. As shown in the figure, the motion vector detection unit 17 of the prediction processing unit 11 includes a motion search control unit 21, a motion vector search unit 22, and an encoding cost calculation unit 23. The motion vector search unit 22 includes a square error calculation unit 221 and a motion vector determination unit 222.

  The motion search control unit 21 controls the processing of the motion vector search unit 22 and the coding cost calculation unit 23 to detect the most appropriate motion vector searched by the motion vector search unit 22.

  The motion vector search unit 22 performs a process for searching for a motion vector. Specifically, using the cost calculated by the encoding cost calculation unit 23, the image displayed in the block at the standard position of the previous frame image (reference image) of the current frame image is the current frame image. Processing for searching to which block of the (encoding target image) has been performed is performed.

  The square error calculation unit 221 performs a calculation using a square error when calculating the motion vector, so that the block at the reference position of the previous frame image of the current frame image and the block of the current frame image are calculated. Processing to calculate the error between.

  The motion vector determination unit 222 starts from the reference position of the previous frame image of the current frame image so that the cost calculated by the encoding cost calculation unit 23 is reduced based on the calculation result of the square error calculation unit 221. Then, a motion vector based on the position and direction up to the movement position in the block of the current frame image is determined.

  The encoding cost calculation unit 23 determines the distance between the blocks based on the information of the reference position of the previous frame image of the current frame image and the information of the blocks of the current frame image. The value of the encoding cost as the information for calculating is calculated.

[Calculation of motion vector]
Hereinafter, the principle of motion vector calculation in this embodiment will be described with reference to the schematic diagram of FIG. In the moving picture compression apparatus 10 of this embodiment, the motion vector is calculated based on this principle.

  For example, as shown in FIG. 3, a previous frame image immediately before the latest frame image that is the reference image (hereinafter referred to as “immediately preceding frame 31”) and the latest frame image that is the encoding target image (hereinafter referred to as “encoding target image”). Is divided into a matrix of a predetermined number of rectangular blocks (for example, m rows × n columns (where m> 1, n> 1)) and approximates the blocks of the respective frame images. Consider a case where a motion vector is calculated by comparing degrees. In this case, a block of the current frame image whose image information is closest to a specific block of the previous previous frame 31, for example, the block 311 is selected, and a motion vector is determined based on the distance and direction of both blocks.

ただし、
ＳＳＤ：Sum of Squared Difference（二乗誤差）
Σ：ブロック内の全ての画素の値の総和
Ｒｄｓｔ：符号化対象フレームの特定の画素の色情報としてのＲ情報
Ｒｓｒｃ：参照フレームの特定の画素の色情報としてのＲ情報
Ｇｄｓｔ：符号化対象フレームの特定の画素の色情報としてのＧ情報
Ｇｓｒｃ：参照フレームの特定の画素の色情報としてのＧ情報
Ｂｄｓｔ：符号化対象フレームの特定の画素の色情報としてのＢ情報
Ｂｓｒｃ：参照フレームの特定の画素の色情報としてのＢ情報
＾：乗算記号。たとえば「○○＾２」は「○○の２乗」を示す。 When the motion vector is calculated in this way, it is generally calculated by an equation using a square error as shown in the following equation (1) (note that this equation has color information “R” “G” in the RGB color space. ”And“ B ”).

However,
SSD: Sum of Squared Difference
Σ: Sum of values of all pixels in the block Rdst: R information as color information of specific pixels of the encoding target frame Rsrc: R information as color information of specific pixels of the reference frame Gdst: Encoding target frame G information as color information of a specific pixel of the G information Gsrc: G information as color information of a specific pixel of the reference frame Bdst: B information as color information of a specific pixel of the encoding target frame Bsrc: Specific information of the reference frame B information ^ as pixel color information: multiplication symbol. For example, “XX ^ 2” indicates “square of XX”.

ただし、
ＳＳＤ：Sum of Squared Difference（二乗誤差）
Σ：ブロック内の全ての画素の値の総和
Ａｄｓｔ：符号化対象フレームの特定の画素の透明度情報（α値の情報）
Ａｓｒｃ：参照フレームの特定の画素の透明度情報（α値の情報）
Ｒｄｓｔ：符号化対象フレームの特定の画素の色情報としてのＲ情報
Ｒｓｒｃ：参照フレームの特定の画素の色情報としてのＲ情報
Ｇｄｓｔ：符号化対象フレームの特定の画素の色情報としてのＧ情報
Ｇｓｒｃ：参照フレームの特定の画素の色情報としてのＧ情報
Ｂｄｓｔ：符号化対象フレームの特定の画素の色情報としてのＢ情報
Ｂｓｒｃ：参照フレームの特定の画素の色情報としてのＢ情報
＾：乗算記号。たとえば「○○＾２」は「○○の２乗」を示す。 In addition, when the pixel information includes transparency information, it is conceivable to simply use the transparency information for each pixel as one element for calculating the square error. In this case, the following color (2) is shown. It becomes an expression like this.

However,
SSD: Sum of Squared Difference
Σ: Sum of values of all pixels in the block Adst: Transparency information of specific pixels of the encoding target frame (α value information)
Asrc: Transparency information (alpha value information) of a specific pixel in the reference frame
Rdst: R information as color information of specific pixels of the encoding target frame Rsrc: R information as color information of specific pixels of the reference frame Gdst: G information Gsrc as color information of specific pixels of the encoding target frame : G information as color information of specific pixel of reference frame Bdst: B information as color information of specific pixel of encoding target frame Bsrc: B information as color information of specific pixel of reference frame ^: multiplication symbol . For example, “XX ^ 2” indicates “square of XX”.

ただし、
ＳＳＤ’：Sum of Squared Difference（二乗誤差）
Σ：ブロック内の全ての画素の値の総和
Ａｄｓｔ：符号化対象フレームの特定の画素の透明度情報（α値の情報）
Ａｓｒｃ：参照フレームの特定の画素の透明度情報（α値の情報、ただし０≦Ａｓｒｃ≦２５５）
Ｒｄｓｔ：符号化対象フレームの特定の画素の色情報としてのＲ情報
Ｒｓｒｃ：参照フレームの特定の画素の色情報としてのＲ情報
Ｇｄｓｔ：符号化対象フレームの特定の画素の色情報としてのＧ情報
Ｇｓｒｃ：参照フレームの特定の画素の色情報としてのＧ情報
Ｂｄｓｔ：符号化対象フレームの特定の画素の色情報としてのＢ情報
Ｂｓｒｃ：参照フレームの特定の画素の色情報としてのＢ情報
＾：乗算記号。たとえば「○○＾２」は「○○の２乗」を示す。 When searching for the most approximate block in the encoding target frame, a cost value obtained by adding a value related to the magnitude of the motion vector to the SSD value calculated by the above equation (2) may be used. Many. The cost value (hereinafter referred to as “cost value”) is obtained by, for example, the following equation (3). Then, the block having the minimum cost value is detected as the start point and end point of the motion vector.
Cost value = SSD + λ consumed motion vector bit (3)
However, λ: a predetermined coefficient for aligning the unit of the number of bits to the unit of SSD In the above formula (2), the transparency information of the pixel is used when calculating the square error, but the transparency information is further utilized. Therefore, in this embodiment, the motion vector determination unit 222 of the motion vector search unit 22 obtains an approximate value between blocks by the following equation (4).

However,
SSD ': Sum of Squared Difference
Σ: Sum of values of all pixels in the block Adst: Transparency information of specific pixels of the encoding target frame (α value information)
Asrc: Transparency information of a specific pixel of the reference frame (alpha value information, where 0 ≦ Asrc ≦ 255)
Rdst: R information as color information of specific pixels of the encoding target frame Rsrc: R information as color information of specific pixels of the reference frame Gdst: G information Gsrc as color information of specific pixels of the encoding target frame : G information as color information of specific pixel of reference frame Bdst: B information as color information of specific pixel of encoding target frame Bsrc: B information as color information of specific pixel of reference frame ^: multiplication symbol . For example, “XX ^ 2” indicates “square of XX”.

Then, the coding cost calculation unit 23 uses “SSD ′” calculated by the above equation (4) instead of “SSD” in the above equation (3), and the cost value of the following equation (5) is minimized. Blocks are detected as the start and end points of motion vectors.
Cost value = SSD ′ + λ consumption vector consumption bit (5)
However, λ: a predetermined coefficient for aligning the unit of the number of bits with the unit of SSD ′ In the above equation (4), “Arc / 255 (where 0 ≦ (Asrc ≦ 255) ”is different from the above formula (2).

  The above equation (4) corresponds to the case where the transparency is set for each pixel in a range of 0 ≦ Asrc ≦ 255 (0: completely transparent state, 255: completely opaque state). For example, if a specific pixel is completely transparent (that is, Asrc = 0), the values of the terms “R”, “G”, and “B” are all “0”. Also, for example, if a specific pixel is translucent (that is, 0 <Arc <255), the values of the terms “R”, “G”, and “B” are the values when completely opaque (that is, when Asrc = 255). Smaller than. And when a specific pixel is translucent, the value of the term of "R" "G" "B" becomes small, so that transparency is high. That is, when determining a motion vector, an error in transparency is more important in determining a motion vector than an error in color information between contrasting pixels or blocks. In many cases, the amount of code at the time of encoding becomes smaller when the error of the color information is ignored or the ratio of the error of the color information to the entire value is reduced.

  Here, the higher the transparency of the image, the lower the importance of the color information in specifying the position on the image. However, in the conventional calculation in the above equation (2), even when the image information of the pixels included in the block is completely transparent or highly transparent, “R”, “G”, “B” The value of “R”, “G” and “B” occupy a high ratio in the entire value obtained by the equation (2). In this case, since color information with low importance cannot be ignored or the ratio of the entire value in the entire value cannot be reduced, reducing the amount of code at the time of encoding is prevented.

  On the other hand, in the calculation of the above formula (4), the higher the transparency, the smaller the ratio of the values of the terms “R”, “G”, and “B” to the entire value of “SSD ′”. That is, the value obtained by the calculation of Expression (4) reflects the importance of transparency more than color information when specifying position information as the transparency of the image is higher. , B contributes to the degree of contribution due to the image quality. Since the calculation result of Expression (4) can ignore the error of the color information or reduce the ratio of the error of the color information in the entire value, the code amount when encoding is small. It becomes small and it becomes possible to make a compression rate high.

[Correlation between color information and transparency]
FIG. 4 is a diagram schematically showing an example of the correlation between color information and transparency in the moving picture compression apparatus 10 of this embodiment. In this figure, the horizontal axis is transparency (α value, where 0 ≦ α value ≦ 255), and the vertical axis is color information contribution or image quality contribution (RGB contribution or YUV contribution, where 0% ≦ contribution Degree ≦ 100%).

  As shown in the figure, in the square error calculation unit 221 of the moving image compression apparatus 10 of this embodiment, color information when searching for a motion vector is detected by detecting the motion vector based on the above equation (3). As shown in the first function 101 shown in FIG. 4, when the α value is 0/255 (that is, when the transparency is 100% transparency), the contribution degree of the color information is 0, α When the value is 127/255 (that is, when the transparency is 50%), the contribution of the color information is 50%, and when the α value is 255/255 (that is, when the transparency is 0%, it is completely opaque), the contribution of the color information is 100%. become. That is, the α value and the color information are in a proportional relationship.

  However, in this embodiment, the calculation in the square error calculation unit 221 of the moving image compression apparatus 10 may not be a complete proportional relationship. For example, as in the second function 102, the third function 103, and the eighth function 108 shown in FIG. 4, the degree of contribution of the color information when the α value is 0/255 is 0. The contribution of color information with an α value of 255/255 is 100%, and there is a general correlation in the middle (when 0/255 <α value <255/255, 0% <color information <100%). In this state, the transparency and the color information may be set to change, and when the α value is 0/255 or more and less than 127/255 as in the fourth function 104, the contribution of the color information is 0. %, When the α value is 127/255 or more and 255/255 or less, the contribution of color information may be set to be 100%.

  In this embodiment, the calculation in the square error calculation unit 221 of the moving image compression apparatus 10 may be a value other than 0/255 or 0% for the minimum value of transparency and color information, or a value of 255/255 for the maximum value. A value other than 100% may be used. For example, as in the fifth function 105 shown in FIG. 4, the maximum value of the color information is set to be less than 100% in a state where the transparency and the color information have a general correlation. Also good. Further, as in the sixth function 106, the minimum value of the color information may be set to a value higher than 0% in a state where the transparency and the color information have a general correlation. Also, as in the seventh function 107, in a state where the transparency and the color information have a general correlation, the minimum value of the color information is a value above 0% and the maximum value is a value less than 100. It may be set as follows. Further, any function other than the first function 101 to the seventh function 107 may be set similarly to these functions. In other words, in the range of the α value used in the square error calculation (0 to 255 in the present embodiment), if the relationship between the α value and the contribution degree of the color information has a tendency to increase, the α value Thus, the motion vector detection is performed in consideration of the above, and the amount of codes can be reduced. Note that “rising upward” is a monotonically increasing function such as the functions 101, 103, 105, 106, and 107, an increasing function such as the functions 102 and 108, or a function 104. Also included are those in which the color information contribution abruptly changes at a predetermined α value.

[Processing procedure]
FIG. 5 is a flowchart illustrating an example of a processing procedure of the moving image compression apparatus 10 according to this embodiment. Hereinafter, an example of the processing procedure of this embodiment as a “moving image data compression method” will be described with reference to FIG. 5 and FIGS. 1 to 4. Note that the processing procedure of FIG. 5 describes processing for searching for a motion vector based on so-called “diamond search”. However, the processing procedure is not limited to this. For example, all blocks of the processing target frame are brute-forced. The motion vector search may be performed by any other processing method such as “full search” for searching.

  The motion search control unit 21 of the motion vector detection unit 17 controls the motion vector search unit 22 to start a motion vector search. The motion vector determination unit 222 of the motion vector search unit 22 uses a matrix of a predetermined number of rectangular blocks (for example, m rows × n columns (where m>) as the immediately preceding frame 31 (hereinafter referred to as “preceding frame”) as a reference image. 1, n> 1)), and a block at a predetermined position (for example, the lower right block 311) is searched for as a reference position.

  Next, the motion vector determination unit 222 divides the latest frame 32 as an encoding target image into a matrix of the same rectangular blocks as the immediately preceding frame 31 (for example, a matrix of m rows × n columns), and immediately preceding the frame 31 of the latest frame 32. The same position (for example, the block 321 at the bottom right) is searched. Then, the motion vector determination unit 222 determines an initial motion vector that uses the searched blocks 311 and 321 of the immediately preceding frame 31 and the latest frame 32 as the start point and end point of the vector.

  Then, the motion vector determination unit 222 includes image information (each of a plurality of pixels constituting each block) from the block 311 of the immediately preceding frame 31 and the block 321 of the latest frame 32 which are the start point and end point of the initial motion vector. Color information (R, G, B values) and transparency information (A values)) are extracted. The extracted information is sent to the square error calculation unit 221, and the square error calculation unit 221 performs calculation by substituting the acquired information into the above equation (3). This processing and calculation in the motion vector determination unit 222 and the square error calculation unit 221 are repeatedly performed for each of the pixels constituting the block 311 of the immediately preceding frame 31 and the block 321 of the latest frame 32 that are the start and end points of the initial motion vector, The sum is calculated as the value of “SSD ′” in equation (3).

  Then, the motion vector search unit 22 sends the calculated “SSD ′” value to the encoding cost calculation unit 23. The motion vector search unit 22 substitutes the acquired value into the above equation (4), and calculates the value of “cost” (step S1, motion vector detection procedure).

  The motion vector search unit 22 sends the calculated “cost” value to the motion vector search unit 22. The motion vector search unit 22 temporarily stores the acquired “cost” value.

  Next, the motion vector search unit 22 calculates the cost of moving the motion vector up, down, left, and right (step S2, motion vector detection procedure). Specifically, the motion vector determining unit 222 (one with the block 311 at the reference position of the immediately preceding frame 31 fixed) and the block 322 immediately above and immediately below the block 321 determined in the latest frame 32 The adjacent block (not present in FIG. 5), the left adjacent block 323, and the right adjacent block (not present in FIG. 5) are respectively searched, and the block 311 at the reference position of the immediately preceding frame 31 is searched. And motion vectors having the searched blocks 322 and 323 as both ends are determined. Then, the square error calculation unit 221 calculates the value of “SSD ′” using Equation (3) for the blocks 311, 322, and 323 at both ends of the determined motion vector. Furthermore, the encoding cost calculation unit 23 calculates the value of “cost” by substituting the value of “SSD ′” into Expression (4). The motion vector determination unit 222 stores the calculated “cost” values.

  Then, the motion vector determination unit 222 compares the “cost” values determined by the blocks 322 and 323 after the block 321 of the latest frame 32 is moved up, down, left, and right, and among them, the “cost” value Is detected as the direction in which the block 321 of the latest frame 32 is moved (step S3, motion vector detection procedure). For example, when the “cost” of the block 322 which is one block higher than the original block 321 in the blocks 322 and 323 of the latest frame 32 after being moved is the smallest, the motion vector determination unit 222 The direction of the next adjacent block 322, that is, the upward direction, is adopted as the moving direction of the block 321 in the latest frame 32.

Then, the motion vector determination unit 222 compares the “cost” of the initial motion vector with the minimum “cost” of the moved block 322 used for adopting the moving direction in step S3. When the minimum “cost” after the movement is smaller than the “cost” of the initial motion vector (“NO” in step S4), the block 322 in the movement direction is set as a new reference block of the latest frame 32. Then, the “cost” of the motion vector between the “cost” of the block 322 and the upper, lower, left and right blocks of the block 322 (the upper adjacent block 324, the left adjacent block 325, and the lower adjacent block 321) is calculated ( In steps S2 and S3, the comparison of costs is repeated while the “cost” of the block 322 after moving from the “cost” of the block 321 is smaller (“NO” in step S4).
Motion vector detection procedure). On the other hand, if the “cost” of the initial motion vector is smaller than the minimum “cost” of the moved block 322 (“YES” in step S4), the motion vector determination unit 222 determines that the “cost” in step S3. ”Is not performed, the block 321 before being moved is detected as the end point (or start point) of the motion vector, and the process ends.

  The motion vector detection unit 17 performs steps S1 to S4 while sequentially setting each frame of the immediately preceding frame 31 as a reference position. When this processing is completed, the prediction processing unit 11 transmits the generated predicted image after the motion compensation processing in the motion compensation unit 18. The moving image compression apparatus 10 generates a prediction error based on the predicted image and the input image, and performs DCT processing in the DCT unit 12. The DCT data generated by the DCT unit 12 is supplied to the code generation unit 13, quantized data is generated by the processing in the quantization unit 19 of the code generation unit 13, and encoded data is generated in the variable length coding unit 20. Thus, compression encoding of the data is performed (compression procedure).

  As described above, in this embodiment, the motion vector between the specific moving image data and the other moving image data is detected using the specific moving image data and / or the transparency information of the other moving image data. Thus, in the moving image data in which the transparency information is set, the motion vector can be detected in a form depending on the set transparency information, and the moving image data can be compressed. Since the motion vector can be detected depending on the transparency information, color information of low importance in detecting the transparent portion in the detection and encoding of the motion vector of the moving image data having the transparent portion is the motion vector. The degree of reflection in detection and encoding can be reduced. Thereby, the compression rate of the moving image containing transparency information can be improved.

  In this embodiment, the motion vector is corrected by correcting the magnitude of the vector formed by the position information of the specific moving image data and the position information of the other moving image data, depending on the value of the transparency level. By detecting, the value of the transparency level is reflected in the detection and encoding of the motion vector, and the compression rate of the moving image including the transparency information can be improved.

  In this embodiment, the higher the transparency defined by the transparency information of the moving image data, the less the influence of the color information of the moving image data on the size of the motion vector, thereby increasing the transparency of the moving image data. Thus, the lower the importance of the color information when detecting the motion vector, the lower the degree to which the color information is reflected in the detection and encoding of the motion vector. Thereby, in the compression of a moving image including transparency information, it is possible to improve the compression rate while reducing less important information.

  In this embodiment, since the motion vector can be detected by correcting the value obtained based on the color information of the moving image data based on the transparency information, in compression of the moving image including the transparency information, The degree of transparency can be reflected in a specific calculation process in motion vector detection and encoding. Accordingly, it is possible to realize in a specific calculation process that the compression rate is improved while reducing the data with low importance based on the degree of transparency.

  In this embodiment, by detecting a motion vector by an absolute value error or a square error of a value obtained by multiplying color information by transparency information, high-precision arithmetic processing based on a value reflecting transparency information And an appropriate motion vector can be detected.

  In this embodiment, by detecting the motion vector using the transparency information, it is possible to perform a highly accurate calculation process based on the value reflecting the transparency information and detect an appropriate motion vector. .

  In this embodiment, the latest frame is the encoding target image and the immediately preceding frame is the reference image. However, the present invention is not limited to this, and the past specific frame is the encoding target image, one of the specific frames. Processing may be performed using a later (time-series later) frame as a reference image. Also, not only the frame before or after the encoding target frame, but also multiple frames before or after multiple frames (for example, two frames before or two frames after a specific frame) are processed as reference images. Alternatively, processing may be performed using a frame before and after a frame of a predetermined time as a reference image.

  The above embodiment is an exemplification of the present invention, and it is needless to say that the present invention is not limited to the above embodiment.

10... Video compression device (video data compression device)
11: Code generator (compression means)
17... Motion vector detection unit (motion vector detection means)

Claims (7)

A video data compression device that compresses video data configured using frame image data in which transparency information is set at least in part,
Motion vector detection means for detecting a motion vector between a specific frame image and other frame image data existing in the forward and / or backward direction of the specific frame image;
Compression means for compressing the frame image data using the detected motion vector,
The motion vector detection means uses the transparency information of the specific frame image data and / or the other frame image data to provide a gap between the specific frame image data and the other frame image data. A motion data compressing device, wherein the motion vector is detected.   The motion vector detection means is configured to determine an image quality in determining a motion vector depending on a value of the transparency of the transparency information of the specific frame image data and / or the other frame image data. The moving image data compression apparatus according to claim 1, wherein the motion vector is detected by correcting a degree of contribution.   When the motion vector is detected, the motion vector detection means defines the transparency specified by the transparency information of the specific frame image data and / or the transparency information of the other frame image data. The higher the transparency, the smaller the influence of the color information of the specific frame image data and / or the color information of the other frame image data on the determination of the motion vector. The moving image data compression apparatus according to claim 2.   The motion vector detection means is a value obtained by multiplying the color information of the specific frame image data by the transparency information of the specific frame image data and / or the color information of the other frame image data. 4. The moving image data compression apparatus according to claim 1, wherein the motion vector is detected using a value obtained by multiplying the transparency information of the frame image data.   The motion vector detection means is a value obtained by multiplying the color information of the specific frame image data by the transparency information of the specific frame image data and / or the color information of the other frame image data. 5. The moving image data compression apparatus according to claim 4, wherein the motion vector is detected using an absolute value error or a square error of a value obtained by multiplying the transparency information of the moving image data.   6. The motion vector detection means detects the motion vector using transparency information of the specific frame image data and / or transparency information of the other frame image data. The apparatus for compressing moving image data according to any one of the above. A method for compressing moving image data for compressing moving image data configured using frame image data in which transparency information is set at least in part,
A motion vector detection procedure in which a motion vector is detected between specific frame image data and other frame image data existing in the forward and / or backward direction of the specific frame image data;
A compression procedure in which the frame image data is compressed using the detected motion vector;
In the motion vector detection procedure, using the transparency information of the specific frame image data and / or the other frame image data, the specific frame image data and the other frame image data are A method for compressing moving image data, wherein the motion vector is detected in between.

Images