JP2008098980A - Moving image encoding apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a picture signal encoding apparatus in which picture quality of a moving image is equalized as well as the processing amount is significantly reduced in comparison with a conventional apparatus. <P>SOLUTION: The moving image encoding apparatus extracts character information of the moving image in encoding an input moving image, and determines a GOP structure based thereon. The apparatus determines a first encoding parameter based on the character information of the moving image and the GOP structure, and calculates encoding complexity for each picture using the character information of the moving image and the GOP structure and the first encoding parameter (a first encoding section 20). Then, the apparatus stores the first encoding parameter and the encoding complexity in a recording medium 30. Next, the apparatus determines a second encoding parameter using the first encoding parameter and the encoding complexity stored in the recording medium 30 in encoding the input moving image of the same content as a previous image again, and encodes the input moving image using the second encoding parameter (a second encoding section 40). <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a moving picture coding apparatus that performs coding using motion compensation prediction.

  2. Description of the Related Art There are known two-pass variable bit rate video signal encoding apparatuses that encode a video signal by dividing it into a first pass and a second pass (for example, JP-A-7-284097 and JP-A-8-265747). Issue gazette). In these conventional video signal encoding devices, for example, after the input two-hour video signal is quantized with a constant quantization width regardless of the amount of generated data in the first pass encoding. Then, variable length coding is performed, and information of the obtained encoded data amount is recorded in the storage circuit for each predetermined unit. When a compression encoding method based on MPEG is adopted as the encoding method, the video signal is recorded in the storage circuit as information on the generated code amount for each video sequence as a picture unit. In JP-A-8-265747, a circuit for determining the GOP structure is added before the first pass. When encoding is performed in the first pass with a fixed quantization width, encoding schemes such as MPEG, which perform variable length encoding, the complexity of the encoded image and the amount of motion compensation residual component are as follows. Accordingly, the amount of generated code increases. Using this property, the distribution state of the generated code amount is investigated in advance.

Then, in the second pass encoding, the code amount is distributed so as to be as close as possible to the distribution, so that the image quality of all video sequences can be made uniform.
In this case, in general, encoding is performed with a smaller quantization width in the first pass, and a larger code amount than the final code amount output in the second pass is generally generated. . This is because by reducing the quantization width, it is necessary to detect fine information up to the high frequency component of the image and to detect the characteristics of the image. Using the almost inversely proportional relationship between the generated code amount and the quantization width, the code amount for each target picture is calculated in the second pass and controlled to that value.

  The two-pass variable bit rate video signal encoding apparatus disclosed in Japanese Patent Application Laid-Open No. 11-275577 has a problem that the above-described conventional example estimates the code amount in units of pictures and cannot reproduce code amount fluctuations in units of MB. It is for solving the problem. In the first pass encoding, the generated code amount in MB units is recorded in the storage circuit as information, and in the second pass encoding, the code amount control is performed in MB units to improve accuracy.

On the other hand, a one-pass variable bit rate video signal encoding device for obtaining encoding efficiency close to that of a two-pass variable bit rate video signal encoding device is known (for example, Japanese Patent Laid-Open No. 10-66067, No. 2002-199398). In this video signal encoding device, the two-pass variable bit rate video signal encoding device uses the characteristic information of the video signal in the previous stage of encoding instead of the generated code amount obtained by the first pass encoding. Thus, it is possible to perform variable bit rate control by calculating the encoding difficulty level from the above and assigning a target code amount according to the encoding difficulty level. In such a one-pass variable bit rate video signal encoding apparatus, it is theoretically difficult to make the image quality uniform for the entire video sequence, but the image quality of the video sequence is made uniform periodically or every scene break. It is possible.
JP-A-7-284097 JP-A-8-265747 Japanese Patent Application Laid-Open No. 11-275577 Japanese Patent Laid-Open No. 10-66067 JP 2002-199398 A

  There is MPEG-4AVC (ISO / IEC14496-10) as a newer encoding method than MPEG-2. In MPEG-4AVC, selectable encoding parameters are prepared several hundred times that of MPEG-2, and by selecting the optimum parameters from among them, the encoding efficiency can be improved more than twice that of MPEG-2. It becomes possible. The encoding parameter is, for example, an MB structure (field / frame), an intra-MB block size, an MB prediction method, a motion vector, or the like.

  Therefore, in MPEG-4 AVC, whether or not an optimal encoding parameter can be selected from a large number of encoding parameter candidates is important for improving the encoding efficiency. It should be noted that the optimal encoding parameter varies with the bit rate (mainly controlled by the quantization width). For example, in general, if the bit rate is high, the selection rate of the macrobook (hereinafter referred to as intra MB) used for intra-frame coding may be increased, but the selection rate of intra MB needs to be decreased at a low bit rate. There is. In MPEG-4 AVC, since the compression efficiency is improved by utilizing the correlation of coding parameters between adjacent MBs, the optimum coding parameter varies depending on the adjacent MB.

Here, consider a case where MPEG-4 AVC is applied to the above-described conventional two-pass variable bit rate video signal encoding apparatus.
In the conventional two-pass variable bit rate video signal encoding apparatus, the code amount (mainly depending on the quantization width) related to the selection of the encoding parameter is not considered, and the first pass, the second pass, Different quantization widths are used. If the first pass and the second pass have similar quantization widths of pictures and MBs, this is not a problem, but the pictures and MBs whose quantization widths of the first pass and the second pass are significantly different In the MB, the code amount estimated in the first pass is significantly different from the actual code amount generated in the second pass, and the optimal encoding parameter changes between the first pass and the second pass. . As described above, in the conventional example, since the variation due to the quantization width of the encoding parameter is not considered, the estimated code amount in the first pass does not work effectively, and a large number of encodings are performed as in MPEG-4 AVC. In an encoding method having parameters, an error in the estimated code amount becomes large. In general, the quantization widths of the first pass and the second pass are close to each other in an average image, but the quantization widths of the first pass and the second pass are greatly different in difficult or simple images. . The variable bit rate technology is a technology for making the image quality of the entire video uniform by improving the coding efficiency for difficult video and simple video, and there is a problem that the characteristics of the original variable bit rate cannot be fully utilized. is there.

Further, the conventional two-pass variable bit rate video signal encoding device does not describe any encoding considering the correlation between adjacent MBs.
As described above, when an encoding method having a large number of encoding parameters such as MPEG-4AVC is applied to the conventional two-pass variable bit rate video signal encoding apparatus, it seems that difficult video and simple video are mixed. Therefore, an error between the code amount estimated by the first pass encoding and the code amount actually generated by the second pass encoding becomes large, and optimal code amount control is performed. Therefore, there is a problem that the image quality of the encoded moving image cannot be made uniform.

  In addition, since MPEG-4 AVC can have a plurality of reference images in one-sided prediction, the processing amount required for motion vector detection is significantly increased compared to MPEG-2. Processing for selecting an optimal mode from a large number of encoding parameters generally requires a huge amount of processing, and there is a problem that it is difficult to achieve both high image quality and reduction in processing amount. For example, in the rate distortion optimization method introduced in the MPEG-4 AVC reference software (ISO / IEC14496-5), since all modes are evaluated after encoding and decoding, the amount of processing is extremely high. Will become bigger.

  As in the conventional two-pass variable bit rate video signal encoding apparatus, selection of encoding parameters with a large amount of processing in both the first pass and the second pass is not from the viewpoint of processing amount. It can be said that it is efficient.

  On the other hand, in the conventional one-pass variable bit rate video signal encoding device, since the encoding difficulty level is calculated from the characteristics of the moving image, the code amount is estimated between the first pass and the second pass. Image quality degradation due to errors does not occur. However, there is a problem regarding the accuracy of selecting an optimal encoding parameter from among hundreds of encoding parameters. There is also a problem in making the image quality uniform for the entire video sequence. Furthermore, no contrivance has been made for reducing processing between the first pass and the second pass.

  Since a higher processing amount is required to perform higher-level encoding than video encoding, the conventional video signal encoding apparatus has been realized by dedicated hardware. However, recently, CPUs have been increased in speed and multi-core, and it has become possible to realize a video signal encoding device by distributed processing using software.

The present invention has been made in view of the above problems.
An encoding parameter having a small influence on the quantization width or an encoding parameter having a low correlation with the adjacent MB is determined at the time of encoding in the first pass, and the encoding parameter having a large influence on the quantization width or an adjacent MB is determined. An encoding parameter with high correlation is determined at the time of encoding in the second pass, thereby solving the problem relating to the dependency of the quantization width in the selection of the encoding parameter, and improving the video quality of the encoded moving image. An object of the present invention is to provide a two-pass variable bit rate video signal encoding device capable of realizing equalization.

  Also, by distributing the processing amount required for encoding in the first pass and the second pass, the processing amount can be significantly reduced as compared with the conventional two-pass variable bit rate video signal encoding device. An object of the present invention is to provide a two-pass variable bit rate video signal encoding device.

Therefore, in order to solve the above problems, the present invention provides the following apparatus.
(1) When encoding an input moving image, among scene change information indicating a scene switching position, fade information indicating a position where a scene fades in and out, and activity information indicating complexity of movement between images An extraction unit for extracting feature information of a moving image including at least one of them,
A first determination unit that determines a GOP structure based on feature information of the moving image;
A second determination unit that determines a first encoding parameter based on the feature information of the moving image and the GOP structure;
A calculation unit that calculates coding complexity for each picture using the feature information of the moving image, the GOP structure, and the first coding parameter;
A storage unit for storing the first encoding parameter and the encoding complexity;
First encoding means comprising:
When the input moving image encoded by the first encoding unit is encoded again, the first encoding unit stores the first moving image stored in the storage unit when the first encoding unit encodes the input moving image. A third determination unit for determining a second encoding parameter using the encoding parameter and the encoding complexity;
Second encoding means for encoding the input moving image using the second encoding parameter;
A moving picture encoding apparatus comprising:
(2) The moving picture coding apparatus according to (1), wherein the first coding parameters are a coding type and a picture structure for each picture.
(3) The moving picture coding apparatus according to (2), wherein an MB structure and a motion vector candidate value are further used as the first coding parameter.

According to the present invention, the following effects can be obtained.
In the first pass encoding, the encoding complexity is calculated from the feature information of the moving image, and the encoding parameter having a small influence of the quantization width and the encoding parameter having a low correlation with the adjacent MB are determined. In the second pass encoding, encoding parameters having a large influence on the quantization width and encoding parameters having high correlation with adjacent MBs are determined and encoded. As a result, it is possible to eliminate the dependency on the quantization width related to the selection of the encoding parameter, so that it is possible to realize a uniform image quality in the entire moving image after encoding.

  In addition, the processing required to select the optimal encoding parameter is performed in a distributed manner between the first pass encoding and the second pass encoding, so that a conventional two-pass variable bit rate video signal is obtained. Compared to the encoding device, the amount of encoding processing can be greatly reduced.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is an overall block diagram showing a configuration of an embodiment of the present invention. FIG. 8 is a flowchart for explaining the overall processing flow in the configuration of FIG.

  The first encoding unit 20 encodes a moving image signal input from a moving image reproducing device such as the VTR 10 and analyzes the characteristics of the moving image in the moving image signal to generate various encoding parameters. To be recorded on the recording medium 30 (step S10).

Next, the second encoding unit 40 records the same moving image signal input from the moving image reproducing device such as the VTR 10 as the one encoded by the first encoding unit 20 on the recording medium 30. The encoding parameter corresponding to the moving image signal is read out, encoding is performed using the encoding parameter, and the obtained bit stream is recorded in the storage medium 50 (step S20).
[First embodiment]
Next, a first example using the above-described embodiment will be described in detail with reference to the drawings.
<First pass>
FIG. 2 is a diagram illustrating an internal configuration of a block that realizes encoding of the first pass in the first embodiment. FIG. 9 is a flowchart showing the flow of the first pass encoding process.

The input image storage unit 100 temporarily stores the input moving image signal (input image signal) (step S100).
Here, the number of images stored in the input image storage unit 100 is NF. Let NF be the maximum possible GOP size Nmax.

  When receiving the next GOP processing start signal from the recording unit 600, the input image storage unit 100 deletes the N images obtained from the GOP structure determination unit 300, and newly stores N images. Depending on the configuration of the moving image feature extraction unit 200, the number of images NF stored in the input image storage unit 100 may be two. Details of these processes will be described later.

The moving image feature extraction unit 200 extracts information indicating the feature of the moving image in the moving image signal supplied from the input image storage unit 100 (step S200).
FIG. 4 is a diagram illustrating a specific configuration example of the moving image feature extraction unit 200. Hereinafter, the operation of the moving image feature extraction unit 200 will be described with reference to FIG.

  The moving image feature extraction unit 200 configures a moving image with a scene change detection unit 210 that detects a scene change point of the moving image, a fade detection unit 220 that detects a scene where the signal level of the moving image gradually changes, and the like. It comprises a motion complexity calculator 230 that calculates the complexity of motion between images, and an activity calculator 240 that calculates the activity of each image.

  The scene change detection unit 210 detects a scene change that is an image switching point in the moving image signal stored in the input image storage unit 100. The absolute difference sum of luminance information between the pixels of each adjacent image, the differential value of the mean square error sum, and the like are calculated. Between the images having the large calculated values is detected as a scene change, and the scene change information a is set to 1. . When no scene change is detected, the scene change information a is set to 0.

  The scene change may be detected using not only the luminance information but also the color difference information in addition to the luminance information. Also, scene changes can be detected after each image has been converted, reduced, etc., or only a specific area within each image can be extracted, and scene changes can be detected using information in these areas. You may do it.

  The fade detection unit 220 detects a scene in which the signal level of the entire image in the moving image signal stored in the input image storage unit 100 varies in a certain direction. The absolute difference sum of the luminance information between the pixels of each adjacent image, the root mean square error sum, and the like are calculated. A scene where the calculated value is a certain value or more is detected as a fade, and the fade information b is set to 1. When no fade is detected, the fade information b is set to 0.

  Note that the fade may be detected using not only the luminance information but also the color difference information in addition to the luminance information. Also, each image can be converted, reduced, etc., to detect fades, or only a specific part of each image can be extracted and fades can be detected using information in these regions. Also good.

  The motion complexity calculation unit 230 calculates the motion between the images in the moving image signal stored in the input image storage unit 100, that is, the motion complexity c. The calculation of the motion complexity c is performed between two images that are temporally adjacent or between two images that have a predetermined time difference. In the calculation of the motion complexity c in the first embodiment, first, one of the two images to be calculated is divided into small blocks of 16 × 16 pixels (referred to as MB), and each MB is divided. Using the luminance information of the pixels in the image, a motion vector search is performed for each MB in comparison with the other image, and a motion vector MV (i) is calculated (i is an MB number). The degree of dispersion of the motion vector of each block obtained as a result is defined as motion complexity c. When these values are large, it indicates that the movement is complicated, while when these values are small, it indicates that the movement is simple.

  The calculation of the motion complexity c does not use the motion vector calculated for each MB as described above, but uses the size of the global motion vector indicating the motion of the entire screen and the amount of temporal change. It is also possible. In addition, a motion vector is calculated as a small block in units of pixels smaller than 16 × 16 MB, such as 16 × 8, 8 × 16, 8 × 8, 8 × 4, 4 × 8, and 4 × 4. It may be used. Further, the calculation of the motion complexity c may be performed using not only luminance information but also color difference information in addition to luminance information. In addition, the motion complexity c is calculated after processing such as conversion and reduction of each image, or only a specific partial area in each image is extracted, and the motion complexity c is used using information in the area. Or may be calculated.

  The activity calculation unit 240 calculates the activity d and the vertical activity e of each image in the moving image signal stored in the input image storage unit 100. The activity d is obtained as a screen internal sum of each MB activity as shown in the following formula 1. Here, NMB is the number of MBs included in the screen. dMB (i) is the MB activity of the i-th MB and is obtained by the following equation 2. Although the absolute difference sum is used in Equations 1 and 2, a square error sum may be used.

・・・式１

... Formula 1

・・・式２

垂直アクティビティｅは、、、、図６及び下記式３に示すように、各ＭＢ内の垂直方向に隣接する各画素間の絶対差分和として求めたＭＢ垂直アクティビティの画面内和として求める。ここで、ｅＭＢ（ｉ）はｉ番目のＭＢのＭＢ垂直アクティビティであり、下記式４で求められる。式３、式４では絶対差分和を用いているが、二乗誤差和を用いても良い。

... Formula 2

The vertical activity e is obtained as an in-screen sum of MB vertical activities obtained as a sum of absolute differences between pixels adjacent in the vertical direction in each MB, as shown in FIG. Here, eMB (i) is the MB vertical activity of the i-th MB and is obtained by the following equation 4. Although the absolute difference sum is used in Equations 3 and 4, a square error sum may be used.

・・・式３

... Formula 3

・・・式４

アクティビティｄ、垂直アクティビティｅの算出は、輝度情報のみでなく、輝度情報に加えて色差情報を用いて行なっても良い。また各画像を変換、縮小等の処理をした後にアクティビティｄ、垂直アクティビティｅを算出したり、各画像内の特定の一部の領域のみを抽出して、この領域内の情報を用いてアクティビティｄ、垂直アクティビティｅを算出したりしても良い。

... Formula 4

The calculation of the activity d and the vertical activity e may be performed using not only luminance information but also color difference information in addition to luminance information. Further, after converting and reducing each image, the activity d and the vertical activity e are calculated, or only a specific partial area in each image is extracted, and the activity d is used by using information in this area. The vertical activity e may be calculated.

Returning to FIG. 2 and FIG.
The moving image feature extraction unit 200 sends the scene change information a, the fade information b, and the motion complexity c of each image detected as described above to the GOP structure determination unit 300. Also, the vertical activity e and the MB vertical activity eMB of each MB are sent to the picture structure determining unit 400. Further, the motion complexity c and activity d are sent to the encoding complexity measuring unit 500.

The GOP structure determination unit 300 determines the coding structure of each image constituting the GOP (step S300).
FIG. 5 is a diagram illustrating a specific configuration example of the GOP structure determination unit 300. Hereinafter, the operation of the GOP structure determination unit 300 will be described with reference to FIG.

The GOP structure determining unit 300 includes a GOP size determining unit 310 that determines the GOP size N and a predicted image interval determining unit 320 that determines the interval M between predicted images.
GOP size determining section 310 determines GOP size N from scene change information a and Nmax. An image from the oldest image stored in the input image storage unit 100 until the scene change information a becomes 1 or Nmax is defined as one GOP. The number of images included in the determined GOP is output as the GOP size N.

  The predicted image interval determination unit 320 uses the GOP size N obtained from the GOP size determination unit 310 and the motion complexity c of the N images obtained from the image feature extraction unit 300 to predict the predicted image interval at the time of encoding the target GOP. Determine M. When the motion complexity c is large, the predicted image interval M is set to a small value, and when the motion complexity c is small, the predicted image interval M is set to a large value. Further, when the motion complexity c is smaller than a value designated in advance, the predicted image interval M can be set as a default value. When the fade information b is 1, the predicted image interval M is 2.

Returning to FIG. 2 and FIG.
The GOP structure determination unit 300 sends the GOP size N and the predicted image interval M determined as described above to the picture structure determination unit 400 and the storage unit 600. Also, the GOP size N is sent to the input image storage unit 100.

  The picture structure determination unit 400 is responsible for part of the encoding parameter determination performed by the MB encoding parameter determination unit 1000 shown in FIG. In the first embodiment, the frame / field structure, which is an inherent characteristic of a moving image among the encoding parameters and has a low dependency on the quantization width, is determined. The picture structure determination unit 400 determines the coding type CT (I image, P image, B image) of each image from the GOP size N obtained from the GOP structure determination unit 300 and the predicted image interval M (step S410). Also, picture structure information PS for encoding each image is determined from the vertical activity e obtained from the image feature extraction unit 300 (step S420). When the picture structure information PS is a frame, the image is encoded with one frame, and when the picture structure information PS is a field, the image is encoded with two fields. When the vertical activity e is small, it is encoded as a frame, and when the vertical activity e is large, it is encoded as a field. Also, the MB structure information MBS of each MB is determined from the MB vertical activity eMB obtained from the image feature extraction unit 300 (step S430). When the MB structure information MBS is a frame, the MB is encoded with one frame, and when the MB structure information MBS is a field, the MB is encoded with two fields. The coding type CT, the picture structure information PS, and the MB structure information MBS determined as described above are collectively used as picture type information PT. Then, this picture type information PT is sent to the encoding complexity measuring unit 500 and the recording unit 600.

  The encoding complexity measurement unit 500 calculates the function of Equation 5 below from the motion complexity c and activity d of N images obtained from the image feature extraction unit 300 and the picture type information PT obtained from the picture structure determination unit 400. Based on f, the encoding complexity C is calculated for each picture (step S500). It is assumed that the function f of the following formula 5 has characteristics as shown in FIG. When the encoded image is an I picture, the encoding complexity C is proportional to the activity d as shown in FIG. When the encoded image is a P picture and a B picture, the encoding complexity C is proportional to the motion complexity c as shown in FIG. In the case of the P picture and the B picture, the function f can consider not only the motion complexity c but also the activity d.

・・・式５

符号化複雑度計測部５００は、上記式５で算出した符号化複雑度Ｃを記録部６００へ送る。

... Formula 5

The encoding complexity measuring unit 500 sends the encoding complexity C calculated by the above equation 5 to the recording unit 600.

The recording unit 600 records the GOP size N obtained from the GOP structure determining unit 300, the picture type information PT obtained from the picture structure determining unit 400, and the coding complexity C obtained from the coding complexity measuring unit 500. (Step S600). Then, a next GOP processing start signal is sent to the input image storage unit 100.
<Second pass>
FIG. 3 is a diagram illustrating an internal configuration of a block that realizes encoding of the second pass in the first embodiment. FIG. 10 is a flowchart showing the flow of the second pass encoding process.

The GOP code amount allocating unit 800 obtains the GOP target code amount TGB from the GOP size N obtained from the recording medium 700 and the GOP encoding complexity (step S700).
The code amount control unit 900 uses the target code amount TGB obtained from the GOP code amount allocating unit 800 and the bit amount code amount B output from the encoding unit 1100, for example, TM5 (Test Code amount control is performed by a code amount control method used in Model 5; ISO / IEC JTC / SC29 1993), and a target code amount TB for each picture is determined (step S800). Then, the target code amount TB is sent to the encoding unit 1100.

The MB encoding parameter determination unit 1000 determines MB encoding parameters necessary for encoding MBs other than the MB structure information MBS, and sends the MB encoding parameters to the encoding unit 1100.
The encoding unit 1100 follows the target code amount TB obtained from the code amount control unit 900, the picture type information PT obtained from the recording medium 700, and the MB coding parameter sent from the MB coding parameter determination unit 1000. An input moving image signal (input image signal) having the same content as that encoded in the first pass is encoded (step S900). The obtained encoded data (bit stream) is output to the outside.
[Second Embodiment]
Next, a second example using the above-described embodiment will be described in detail with reference to the drawings.
<First pass>
FIG. 11 is a diagram illustrating an internal configuration of a block that realizes the first pass encoding in the second embodiment.

  The input image storage unit 100, the GOP structure determination unit 300, the picture structure determination unit 400, the encoding complexity measurement unit 500, and the recording medium 700 have the same functions as those in the first embodiment, and are the same as those in FIG. The code is attached.

In addition to the functions of the first embodiment, the moving image feature extraction unit 201 sends the motion vector information g obtained by the motion complexity calculation unit 230 shown in FIG. 4 to the recording unit 600.
The recording unit 601 records the motion vector information g obtained from the moving image feature extraction unit 201 on the recording medium 700 in addition to the functions of the first embodiment.
<Second pass>
FIG. 12 is a diagram illustrating an internal configuration of a block that realizes encoding of the second pass in the second embodiment. FIG. 10 is a flowchart showing the flow of the second pass encoding process.

The recording medium 700, the GOP code amount assigning unit 800, and the code amount control unit 900 have the same functions as those in the first embodiment, and are given the same reference numerals as those in FIG.
The MB encoding parameter determination unit 1001 obtains motion vector information to be encoded by using the motion vector information g obtained from the recording medium 700 as one motion vector candidate, and further determines MBs other than the MB structure information MBS. MB encoding parameters necessary for encoding are determined and sent to the encoding unit 1101.

  The encoding unit 1101 is based on the target code amount TB obtained from the code amount control unit 900, the picture type information PT obtained from the recording medium 700, and the MB coding parameter sent from the MB coding parameter determination unit 1001. An input moving image signal (input image signal) having the same content as that encoded in the first pass is encoded. The obtained encoded data (bit stream) is output to the outside.

  As in the second embodiment, one candidate for a motion vector is obtained by encoding the first pass, and the motion vector is used at the time of encoding the second pass. Therefore, it is possible to greatly reduce the processing amount of the encoding unit 1101 that performs the motion vector detection process that occupies a large weight.

  The present invention also includes a program for causing a computer to realize the functions of the above-described moving picture coding apparatus. These programs may be read from a recording medium and loaded into a computer, or may be transmitted via a communication network and loaded into a computer.

It is a whole block diagram which shows embodiment of this invention. It is a block diagram which shows the 1st code | symbol part in 1st Example of this invention. It is a block diagram which shows the 2nd code | symbol part in 1st Example of this invention. It is a block diagram which shows the moving image feature extraction part in the Example of this invention. It is a block diagram which shows the GOP size determination part in the Example of this invention. It is a figure explaining the vertical activity in the Example of this invention. It is a figure explaining the calculation method of the complexity in the Example of this invention. It is a flowchart which shows the whole process in embodiment of this invention. It is a flowchart which shows the 1st code | symbol part in the Example of this invention. It is a flowchart which shows the 2nd code | symbol part in the Example of this invention. It is a block diagram which shows the 1st code | symbol part in 2nd Example of this invention. It is a block diagram which shows the 2nd code | symbol part in 2nd Example of this invention.

Explanation of symbols

10 VTR
20 first encoding unit 30 recording medium 40 second encoding unit 50 recording medium 100 input image storage unit 200 moving image feature extraction unit 300 GOP structure determination unit 310 GOP size determination unit 320 predicted image interval determination unit 400 picture structure determination unit 500 Complexity measurement unit 600 recording unit 700 recording medium 800 GOP code amount allocation unit 900 code amount control unit 1000 MB coding parameter determination unit 1100 encoding unit 201 moving image feature extraction unit (second embodiment)
601 Recording unit (second embodiment)
1001 MB coding parameter determination unit (second embodiment)
1101 Code part (second embodiment)

Claims (3)

When encoding an input video, at least one of scene change information indicating the scene switching position, fade information indicating the position where the scene fades in and out, and activity information indicating the complexity of movement between images An extraction unit for extracting feature information of a moving image including one;
A first determination unit that determines a GOP structure based on feature information of the moving image;
A second determination unit that determines a first encoding parameter based on the feature information of the moving image and the GOP structure;
A calculation unit that calculates coding complexity for each picture using the feature information of the moving image, the GOP structure, and the first coding parameter;
A storage unit for storing the first encoding parameter and the encoding complexity;
First encoding means comprising:
When the input moving image encoded by the first encoding unit is encoded again, the first encoding unit stores the first moving image stored in the storage unit when the first encoding unit encodes the input moving image. A third determination unit for determining a second encoding parameter using the encoding parameter and the encoding complexity;
Second encoding means for encoding the input moving image using the second encoding parameter;
A moving picture encoding apparatus comprising:   The moving image encoding apparatus according to claim 1, wherein the second determination unit is a unit that determines an encoding type and a picture structure for each image as the first encoding parameter.   3. The moving picture encoding apparatus according to claim 2, wherein the second determination unit is means for further determining an MB structure and a motion vector candidate value as the first encoding parameter.

Images