JPWO2008053557A1 - Moving image re-encoding device, moving image re-encoding method, moving image re-encoding program, and recording medium storing moving image re-encoding program - Google Patents

Abstract

In a moving image re-encoding apparatus that once decodes an encoded moving image and re-encodes it, a delay time and memory are reduced and an appropriate code amount is predicted. A GOP structure prediction unit in a coding information analysis unit counts, for example, the number of P pictures in eight GOPs from a picture type obtained when an input bit stream is decoded by a decoding unit. An average value of the number of P pictures and the number of B pictures is calculated from six GOPs excluding the GOP having the maximum value and the minimum value, and a predicted GOP structure is obtained. The encoding unit 103 performs re-encoding with reference to the predicted GOP structure. [Selection] Figure 1

Description

  The present invention relates to a moving image re-encoding device, a moving image re-encoding method, a moving image re-encoding program, and a moving image that re-encode a moving image that has been encoded once by changing an encoding method, a bit rate, or the like. The present invention relates to a recording medium storing a re-encoding program.

  In many cases, moving images are digitized and recorded or broadcast. However, since moving images have an enormous amount of data, they are usually recorded or broadcast with a reduced amount of data by encoding. MPEG (Moving Picture Experts Group) -2 or the like is used as such a moving image encoding method.

  A moving picture encoded by MPEG-2 or the like is re-encoded by converting the bit rate according to the capacity of a recording medium to be recorded or the speed of a communication line to be transmitted, or recently introduced MPEG-4 or H.264. In some cases, re-encoding may be performed using a higher-efficiency encoding method than MPEG-2 such as H.264.

For example, there is a transcoder described in Patent Document 1 as an apparatus for once decoding and re-encoding a once-encoded moving image. In the transcoder described in Patent Document 1, the I frame, the P frame, and the B frame that are decoded in the order in which the decoding unit 1 is input, the I frame is the I frame, the P frame is the P frame, and the B frame in the decoding order. Is encoded as a B frame, thereby reducing delay time and memory.
JP 2000-92497 A

  Since the transcoder described in Patent Document 1 described above does not have a rearranger that rearranges images in the display order when decoding, delay time and memory for rearranging the images in the display order can be reduced. However, when re-encoding is performed by changing the bit rate or the encoding method, the code amount allocated at the time of re-encoding is often different from the bit stream before decoding. For example, there is a possibility that the image quality deteriorates because an appropriate code amount is not assigned.

  In order to assign an appropriate code amount, for example, an appropriate code amount can be assigned by storing a certain amount of images such as 1 GOP (Group Of Pictures) in a memory or the like. In addition to the increase in cost due to the need for memory, the delay time increases.

  Therefore, for example, when re-encoding an encoded moving image by changing a bit rate or an encoding method, the present invention reduces the delay time and memory and encodes an appropriate code amount. An object is to provide a moving image re-encoding device, a moving image re-encoding method, a moving image re-encoding program, and a recording medium storing the moving image re-encoding program.

  In order to solve the above-mentioned problem, the invention according to claim 1 is characterized in that at least one intra-coded image is selected from among an intra-coded image, a forward prediction coded image, and a bidirectional predictive coded image. Based on the decoding unit that receives the bit stream composed of the image group included above and decodes the bit stream to obtain the image data and the encoding information, and the image data and the encoding information decoded by the decoding unit And a coding means for coding the image data again, and counting the number of forward prediction coded images in each of the plurality of image groups based on the coding information. And the front of at least one of the image groups having the maximum value or the minimum value of the number of forward prediction encoded images in each of the counted plurality of image groups. An image that outputs an average value of the forward prediction encoded image and the bidirectional predictive encoded image in a plurality of the image groups excluding one or more image groups to the encoding unit as a predicted structure of the image group It has a group structure prediction means.

  The invention according to claim 8 is constituted by an image group including at least one of the intra-coded images among the intra-coded image, the forward predictive coded image, and the bidirectional predictive coded image. In a moving image re-encoding method in which a bit stream is input, the bit stream is decoded, image data and encoding information are obtained, and the image data is encoded again based on the decoded image data and the encoding information The number of forward prediction encoded images in each of the plurality of image groups is counted based on the encoding information, and the maximum number of forward prediction encoded images in each of the plurality of counted image groups is counted. The forward predictive encoded image in a plurality of the image groups excluding at least one image group of at least one of the image groups having a value or a minimum value. The average value of the fine the bidirectional predictive coded picture, are characterized by using during the encoding the structure of the image group predicted.

  The invention according to claim 9 is configured by an image group including at least one of the intra-coded images among the intra-coded image, the forward predictive coded image, and the bidirectional predictive coded image. A decoding unit that receives the bit stream to be decoded and obtains image data and encoding information, and encodes the image data again based on the image data and the encoding information decoded by the decoding unit In a moving image re-encoding program that causes a computer to function as an encoding unit, the number of forward prediction encoded images in each of the plurality of image groups is counted based on the encoding information, and the plurality of counted At least one of the image groups having the maximum value or the minimum value of the number of forward prediction encoded images in each image group. An image that outputs an average value of the forward prediction encoded image and the bidirectional predictive encoded image in a plurality of the image groups excluding one or more image groups to the encoding unit as a predicted structure of the image group It is characterized by having a computer function as group structure prediction means.

It is a block diagram of the moving image re-encoding apparatus concerning one Example of this invention. It is a block diagram of the encoding information analysis part of the moving image re-encoding apparatus shown by FIG. It is explanatory drawing of an example of GOP. It is the flowchart explaining operation | movement of the GOP structure estimation part. It is explanatory drawing of the count of the number of P pictures and the number of B pictures in each GOP. It is explanatory drawing at the time of calculating | requiring the average of P picture and B picture from several GOP. It is the flowchart explaining operation | movement of the encoding frame rate estimation part. It is explanatory drawing of an encoding picture and a display picture. It is explanatory drawing of the count of the encoding picture number in each GOP and the number of display pictures. It is the flowchart explaining operation | movement of the GOP code amount prediction part. It is explanatory drawing at the time of calculating | requiring the average of P picture and B picture in case there exist two or more maximum values or minimum values of P picture. It is an operation | movement flowchart of a moving image encoding program.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Moving image re-encoding apparatus 101 Decoding part (decoding means)
102 Encoding information analysis unit 103 Encoding unit (encoding means)
200 GOP structure prediction unit (image group structure prediction means)
201 Coding frame rate prediction unit (frame rate prediction means)
202 GOP code amount prediction unit (code amount prediction means)

  Hereinafter, a moving image re-encoding device according to an embodiment of the present invention will be described. The moving image re-encoding device according to an embodiment of the present invention includes the number of forward-predictive encoded images and bidirectional predictive-encoded images in an image group for reference by the image group structure predicting unit during re-encoding. Is predicted as the structure of the image group, at least one of the image group having the maximum value or the image group having the minimum value of the number of forward prediction encoded images in the plurality of image groups is one. The average value of the forward prediction encoded images and the average value of the bidirectional predictive encoded images of the plurality of image groups excluded as described above are predicted as the structure of the image group, and the encoding unit re-generates based on the structure of the image group. Encode. By doing so, it is possible to predict the structure of a more average image group without being affected by the image group having the extreme number of forward-predicted encoded images that appear temporarily. A large amount of code can be predicted. Further, since it is not necessary to hold an extra image for the structure prediction of the image group, the delay time and the memory amount are reduced.

  In addition, the image group structure prediction means includes a front of a plurality of image groups obtained by excluding one or more image groups each having a maximum value and a minimum value from the number of forward prediction encoded images in each of the plurality of counted image groups. The average value of the direction prediction encoded image and the bidirectional prediction encoded image may be output to the encoding means as the structure of the image group. By doing in this way, an average image can be obtained with higher accuracy without being affected by an image group having an extreme number of forward-predictive encoded images that temporarily appear in both the maximum value direction and the minimum value direction. The structure of the group can be predicted.

  Further, the image group structure prediction means includes a forward prediction encoded image in a plurality of image groups excluding the plurality of image groups in descending order among the counted number of forward prediction encoded images in each of the plurality of image groups, and You may output the average value of a bi-directional predictive coding image to a coding means as a structure of an image group. By doing in this way, even if there are a plurality of image groups having the number of extreme forward predictive encoded images that temporarily appear in the maximum value direction, it is not affected by them and the average can be obtained with higher accuracy. The structure of the image group can be predicted.

  Further, the image group structure prediction means includes a forward prediction encoded image in a plurality of image groups excluding the plurality of image groups in order from the smallest number of forward prediction encoded images in each of the plurality of counted image groups, and You may output the average value of a bi-directional predictive coding image to a coding means as a structure of an image group. By doing this, even if there are a plurality of image groups having the number of extreme forward predictive encoded images that temporarily appear in the minimum value direction, the average is more accurate without being affected by them. The structure of the image group can be predicted.

  Further, the image group structure prediction means counts the number of bidirectional prediction encoded images in each of the plurality of image groups, and the maximum value or the number of forward prediction encoded images in each of the plurality of image groups counted When there are a plurality of image groups each having a minimum value, among the plurality of image groups having the maximum value or the minimum value, the image group has a value far from the average value of the bi-directional predictive encoded images in the counted plurality of image groups. The average value of the forward prediction encoded image and the bidirectional predictive encoded image in a plurality of image groups excluding the image group may be output to the encoding means as the structure of the image group. By doing this, even if there are a plurality of image groups having the maximum value or the minimum value of the forward prediction encoded image, it is possible to exclude the one having the extreme value of the bidirectional predictive encoded image. The structure of the average image group can be predicted with higher accuracy not only in the direction prediction encoded image but also in the bidirectional prediction encoded image.

  In addition, based on the encoding information, the number of encoded images in each of the plurality of image groups and the number of images to be displayed in each of the plurality of image groups are counted, respectively. From the ratio between the number of images to be displayed and the number of images to be displayed in each of the plurality of image groups, and the encoding information, the number of images (encoding frame rate) to be re-encoded by the encoding unit per second is predicted. You may have a frame rate estimation means. In this way, since not only the number of encoded images but also the number of images to be displayed are referred to, if there are many images having redisplay information (repeat picture information), reencoding is performed. In this case, since a substantial encoding frame rate can be accurately predicted, a more appropriate code amount can be predicted.

  In addition, the encoding unit encodes based on the structure of the image group predicted by the image group structure prediction unit, the encoded frame rate predicted by the frame rate prediction unit, and a preset re-encoding bit rate. Code amount prediction means for predicting the code amount per image group at the time of image processing may be provided. By doing so, it is possible to accurately predict the code amount per image group in consideration of the image group structure and the repeat picture when re-encoding.

  In addition, the moving image re-encoding method according to the embodiment of the present invention uses the number of forward prediction encoded images and bidirectional predictive encoded images in the image group as a structure of the image group for reference during re-encoding. A plurality of images excluding at least one image group of at least one of an image group having a maximum value or an image group having a minimum value of the number of forward prediction encoded images in a plurality of image groups when performing prediction Based on the structure of the image group, re-encoding is performed as the structure of the image group in which the average value of the forward prediction encoded image of the group and the average value of the bidirectional predictive encoded image are predicted. By doing in this way, it is possible to predict a more average image group structure without being influenced by an image group having an extreme number of forward-predictive encoded images that appear temporarily. Further, since there is no need to hold an extra image for the image group prediction, the delay time and the memory amount are reduced.

  In addition, the moving image re-encoding program according to the embodiment of the present invention includes a forward-predictive encoded image and a bi-directional predictive-encoded image in an image group that the image group structure prediction unit refers to when re-encoding. When predicting the number of images as the structure of the image group, at least one of the image group having the maximum value or the image group having the minimum value of the number of forward prediction encoded images in the plurality of image groups is selected. The average value of the forward prediction encoded images and the average value of the bidirectional predictive encoded images of a plurality of image groups excluding one or more are set as the predicted image group structure, and the encoding unit is based on the structure of the image group. And let the computer function to re-encode. By doing in this way, it is possible to predict a more average image group structure without being influenced by an image group having an extreme number of forward-predictive encoded images that appear temporarily. Further, since there is no need to hold an extra image for the image group prediction, the delay time and the memory amount are reduced.

  Further, the moving image re-encoding program according to claim 9 may be stored in a recording medium. In this way, the moving image re-encoding program can be distributed alone as well as being incorporated into the device.

  A moving image re-encoding device 100 according to an embodiment of the present invention will be described with reference to FIGS. The moving image re-encoding device 100 is a device that re-encodes a bit stream obtained by encoding a moving image input from the input terminal 110 while changing the bit rate and the encoding method and outputs the re-encoded image from the output terminal 115. As shown in FIG. 1, the moving image re-encoding apparatus 100 includes a decoding unit 101, an encoded information analysis unit 102, an encoding unit 103, an input terminal 110, a setting data terminal 111, and an output terminal 115. ing.

  The decoding unit 101 as decoding means decodes the bit stream input from the input terminal 110, outputs the image data 113 to the encoding unit 103, and encodes the encoded information 112 into the encoding unit 103 and the encoded information analysis unit. To 102. For example, when a bit stream encoded by the MPEG-2 method is input, the bit stream is decoded, and decoded image data 113 (luminance, color difference data, etc.) is output to the encoding unit 103. Header information and the like in the input bitstream are output to the encoded information analysis unit 102 and the encoding unit 103 as encoded information 112.

  The encoding information 112 is, for example, a picture type (I picture, P picture, or B picture), field redisplay information, frame redisplay information, and frame rate information (frame rate when displayed on the screen, for example, In the case of NTSC, 29.97 frames / second) is output. I picture, P picture, and B picture are classifications of encoded images used in MPEG, etc., I picture is an image that has been subjected to intra-picture encoding, and P picture is an image that has been subjected to forward predictive encoding , B pictures show images that have been subjected to bi-directional predictive coding.

  The encoding information analysis unit 102 includes a GOP structure prediction unit 200, an encoding frame rate prediction unit 201, and a GOP code amount prediction unit 202 as shown in FIG. Setting data and encoding information 112 are input, and based on the setting data and encoding information 112, an allocation code amount when re-encoding is calculated in the encoding unit 103 is calculated, and analysis result information 114 is transmitted to the encoding unit 103. Output.

  The setting data terminal 111 receives parameters set when the moving image re-encoding device 100 is initially set, for example, the bit rate of the output bit stream.

  As the analysis result information 114, for example, the allocated code amount in GOP units, the predicted number of P pictures in the GOP, the predicted number of B pictures, the encoding frame rate, and the like are output. GOP as an image group is an abbreviation of Group Of Pictures, and is a unit composed of at least one I picture (usually composed of a plurality of pictures). The GOP structure is a configuration of the number of I pictures, P pictures, and B pictures in a GOP. FIG. 3 shows an example of the GOP structure. A GOP starts from an I picture, and a GOP is the one picture before the next I picture. In FIG. 3, 1 GOP is composed of 12 pictures. There are 1 I picture, 3 P pictures, and 8 B pictures in the GOP.

  The GOP structure prediction unit 200 as an image group structure prediction unit receives the encoded information 112, predicts the GOP structure based on the encoded information 112, and outputs the GOP structure information 210.

  A GOP structure prediction method in the GOP structure prediction unit 200 will be described with reference to the flowchart of FIG.

  First, in step S101, an initial value of the GOP structure is set, and the process proceeds to step S102. The initial value of the GOP structure is the number of I pictures, P pictures, and B pictures set before moving pictures are input. These are initial values of prediction and do not need to match the number of pictures of the moving image actually input. The initial value may be previously stored in the GOP structure prediction unit 200 or may be set from the setting data terminal 111.

  Next, in step S102, the picture type of the image input to the decoding unit 101 is input via the encoding information 112, and the process proceeds to step S103.

  Next, in step S103, it is determined whether or not the picture type of the image input to the decoding unit 101 is an I picture. If the picture type is an I picture (YES), the process proceeds to step S106; In case), the process proceeds to step S104.

  Next, in step S104, it is determined whether the picture type of the image input to the decoding unit 101 is a P picture. If the picture type is a P picture (in the case of YES), the process proceeds to step S108; If), the process proceeds to step S105.

  Next, in step S105, if it is neither an I picture nor a P picture, it is determined that the picture type of the image input to the decoding unit 101 is a B picture, and the number of B pictures in the current GOP is set to 1. Add and return to step S102.

  In step S106, it is determined that an I picture has been input. Since the range from the I picture to the previous image of the next I picture is in the range of 1 GOP, assuming that the top of the GOP is detected by I picture detection, the GOP having the maximum value and the minimum value of the number of P pictures in the previous plurality of GOPs is 1 The average value of the number of P pictures and the average number of B pictures of a plurality of past GOPs, excluding one by one and removing the maximum and minimum GOPs of the P pictures one by one, proceeds to step S107. That is, the average value of P pictures and B pictures in a plurality of GOPs is calculated by removing one or more GOPs having the maximum value and the minimum value from the number of P pictures in the plurality of GOPs. That is, the average value of the P picture and the B picture in a plurality of GOPs excluding at least one of the GOPs having the minimum value or the maximum value is calculated.

  Step S106 will be described in detail with reference to FIGS. FIG. 5 shows an example in which 8 GOPs are held as a plurality of previous GOPs. Assuming that the GOP including the input I picture is the current GOP, GOP [0], GOP [1], GOP [2], GOP [3], GOP [4], GOP are in order of time from the current GOP. [5], GOP [6], GOP [7], and the number of P pictures in each GOP is Np [0], Np [1], Np [2], Np [3], Np [4], Np [ 5], Np [6], Np [7], and the number of B pictures in each GOP is Nb [0], Nb [1], Nb [2], Nb [3], Nb [4], Nb [5 ], Nb [6], and Nb [7]. That is, the number of P pictures and B pictures for 8 GOPs from the immediately preceding GOP is always held. If the number of P pictures and the number of B pictures in each GOP are as shown in FIG. 6, except for Np [2], which is the maximum value of P pictures, and Np [6], which is the minimum value. When the average value AVR_Np of the P picture and the average value AVR_Nb of the B picture are obtained, values 4 and 10 are obtained as shown in FIG.

  Next, in step S107, the average value (AVR_Np, AVR_Nb) of the P picture and B picture calculated in step S106 is output to the GOP code amount prediction unit 202 as GOP structure information 210, and the process returns to step S102.

  In step S108, since the picture type of the image input to the decoding unit 101 is P picture, 1 is added to the number of P pictures in the current GOP, and the process returns to step S102.

  An encoding frame rate prediction unit 201 as a frame rate prediction unit receives the encoding information 112, calculates an encoding frame rate based on the encoding information 112, and outputs encoded frame rate information 211.

  A method of calculating the encoding frame rate in the encoding frame rate prediction unit 201 will be described with reference to the flowchart of FIG.

  First, in step S201, initial values of the number of displayed pictures and the number of encoded pictures are set, and the process proceeds to step S202. These are initial values of prediction and do not need to match the number of display pictures or the number of encoded pictures that are actually input. The initial value may be stored in advance in the encoding frame rate prediction unit 201 or may be set from the setting data terminal 111.

  The number of displayed pictures and the number of encoded pictures will be described with reference to FIG. The number of displayed pictures as the number of images to be displayed indicates the number of pictures when decoded and displayed on the screen, and the number of encoded pictures as the number of encoded images is actually encoded. It is the number of pictures. The picture for which field redisplay information or frame redisplay information is detected when decoding is performed by the decoding unit 101 has one encoded picture, but the number of displayed pictures is one or more. In the case of FIG. 8, when the number of displayed pictures per second is 12 between the arrows, the frame rate is 12 frames / second and the encoding frame rate is 8 frames / second. The frame rate information included in the encoding information 112 indicates the frame rate of the display picture. Here, when the code amount for 1 second is assigned to 12 encoded pictures from only the frame rate information without detecting redisplay information, the code amount per frame is 1/12 for 1 second. . However, if the re-display information is detected, it can be seen that there are 8 encoded pictures per second, so the allocated code amount per frame increases to 1/8 of a second. Therefore, by using the redisplay information, it is possible to assign an appropriate code by allocating the code amount by using the actual number of encoded pictures excluding the number of repeat pictures at the time of re-encoding as a re-encoding target.

  Next, in step S202, the picture type of the image input to the decoding unit 101 is input via the encoding information 112, and the process proceeds to step S203.

  Next, in step S203, it is determined whether or not the picture type of the image input to the decoding unit 101 is an I picture. If the picture type is an I picture (YES), the process proceeds to step S204; In case), the process proceeds to step S206.

  In step S204, it is determined that an I picture has been input. Since the range from the I picture to the previous image of the next I picture is in the range of 1 GOP, assuming that the top of the GOP is detected by I picture detection, the encoding frame rate is calculated from the number of encoded pictures and the number of displayed pictures in the previous plurality of GOPs. The calculation proceeds to step S205.

  The encoded frame rate is the past i-th counted coded_frame_rate, num is the number of GOPs holding data, i = 0,1, ..., (num-1), Nc [i] is counted immediately before When the number of encoded pictures in the GOP, Nd [i] is the number of displayed pictures in the past i-th GOP, and frame_rate is the frame rate information included in the encoded information 112, for example, the following expression 1 is obtained.

  That is, a predicted encoded frame rate is calculated from the ratio between the number of encoded pictures and the number of displayed pictures and the frame rate information included in the encoded information 112.

  Step S204 will be described in detail with reference to FIG. FIG. 9 shows an example when 8 GOPs are held as a plurality of previous GOPs. Assuming that the GOP including the input I picture is the current GOP, GOP [0], GOP [1], GOP [2], GOP [3], GOP [4], GOP are in order of time from the current GOP. [5], GOP [6], GOP [7], and the number of encoded pictures in each GOP is Nc [0], Nc [1], Nc [2], Nc [3], Nc [4], Nc [ 5], Nc [6], Nc [7], and the number of displayed pictures in each GOP is Nd [0], Nd [1], Nd [2], Nd [3], Nd [4], Nd [5] , Nd [6] and Nd [7]. That is, the number of encoded pictures and the number of displayed pictures are always held for 8 GOPs from the immediately preceding GOP. When these are applied to Equation 1, the encoded frame rate predicted by multiplying the total number from Nc [0] to Nc [7] by the total number from Nd [0] to Nd [7] and the frame rate information. Can be obtained as

  Next, in step S205, the encoded frame rate (coded_frame_rate) calculated in step S204 is output to the GOP code amount prediction unit 202 as encoded frame rate information 211, and the process proceeds to step S206.

  Next, in step S206, 1 or (1 + the number of repeat pictures) is added to the number of encoded pictures and the number of displayed pictures of the current GOP from the encoding information 112, and the process returns to step S202. That is, 1 is always added to the number of encoded pictures, and (1 + number of repeat pictures) is added to the number of displayed pictures.

  The GOP code amount prediction unit 202 serving as a code amount prediction unit performs GOP based on the input encoding information 112, the GOP structure information 210, the encoding frame rate information 211, and the output target bit rate set from the setting data terminal 111. Is allocated to the encoding unit 103 as analysis result information 114 together with the predicted number of P pictures, the predicted number of B pictures included in the GOP structure information 210, and the encoded frame rate included in the encoded frame rate information 211. Output.

  The operation of the GOP code amount prediction unit 202 will be described with reference to the flowchart of FIG.

  First, in step S301, the picture type of the image input to the decoding unit 101 is input via the encoding information 112, and the process proceeds to step S302.

  Next, in step S302, it is determined whether or not the picture type of the image input to the decoding unit 101 is an I picture. If the picture type is an I picture (in the case of YES), the process proceeds to step S303; If), the process returns to step S301.

  Next, in step S303, it is determined whether or not the GOP structure information 210 has been updated. If updated (in the case of YES), the process proceeds to step S304, and if not (in the case of NO), the process waits until it is updated. To do.

  Next, in step S304, it is determined whether or not the encoded frame rate information 211 has been updated. If updated (in the case of YES), the process proceeds to step S305, and if not (in the case of NO), it is updated. Wait until.

  Next, in step S305, the allocated code amount per GOP is calculated based on the encoded information 112, the GOP structure information 210, the encoded frame rate information 211, and the output target bit rate set from the setting data terminal 111, and in step S306. Proceed to

  The allocated code amount GOP_bits per GOP is the average number of P pictures predicted by the GOP structure prediction unit 200 for AVR_Np, the average number of B pictures predicted by the GOP structure prediction unit 200, and the coded_frame_rate is predicted by the encoded frame rate prediction unit 201. Assuming that the encoded frame rate and bitrate are output target bit rates set in advance from the setting data terminal 111, they are expressed by, for example, the equation (2).

  The portion of 1 in the equation shown in Equation 2 means the number of I pictures, and at the time of field coding, since one frame is 2 fields, 1 of the average number of I fields in the GOP. / 2 can be calculated and replaced. (In this case, AVR_Np and AVR_Nb are also calculated by half the number of fields.)

  Next, in step S306, the allocated code amount (GOP_bits) per GOP calculated in step S305, the GOP structure information 210, and the encoded frame rate information 211 are output to the encoding unit 103 as analysis result information 114, and the process proceeds to step S301. Return.

  The encoding unit 103 as an encoding unit re-encodes the image data 113 input from the decoding unit 101 based on the encoding information 112, the analysis result information 114, and the initial setting input from the setting data terminal 111, and outputs the result. The bit stream is output from the output terminal 115. That is, code amount control is performed based on the allocated code amount per GOP predicted by the encoded information analysis unit 102, and the code amount is assigned to each picture based on the predicted number of P pictures and B pictures. Also, the code amount control may be changed based on the encoded information 112, information in the encoding unit 103, and the like. For example, after obtaining the allocated code amount per GOP of the analysis result information 114, the allocated code amount of GOP or picture may be increased or decreased based on the complexity of the input image data 113.

  According to the present embodiment, in the GOP structure prediction unit 200, the average value of the number of P pictures and the number of B pictures from six GOPs excluding the GOP having the maximum value and the minimum value among the number of P pictures in the eight GOPs. And the predicted GOP structure. The encoded frame rate prediction unit 201 calculates the encoded frame rate from the number of encoded pictures, the number of displayed pictures, and the frame rate obtained from the encoded information 112, and sets it as the predicted encoded frame rate. Then, a code amount per GOP is calculated from the predicted GOP structure, the predicted encoding frame rate, and the bit rate set from the setting data terminal 111, and is used as the predicted code amount. The encoding unit 103 performs re-encoding with reference to the predicted GOP structure, the predicted encoding frame rate, and the predicted code amount. By doing so, the prediction accuracy of the GOP structure, the frame rate, and the code amount is improved, and it is not necessary to store a plurality of images in the memory for the GOP structure prediction, the frame rate prediction, and the code amount prediction. Delay time and memory amount can be reduced.

  In the above-described embodiment, the GOP structure prediction unit 200 calculates the average number of GOPs of P and B pictures except for the GOP having the maximum value and the GOP having the minimum value among the number of P pictures for each GOP. However, as shown in FIG. 11, when there are a plurality of maximum values or minimum values of P pictures, the number of B pictures is compared to exclude GOPs that have a large difference from the average value of B pictures. May be. In FIG. 11, since the maximum values of the P picture are Np [2] and Np [3] and the minimum values are Np [6] and Np [7], Nb [2], Nb [3] and Nb The number of B pictures of [6] and Nb [7] is compared, and the GOP including Nb [2] having a large difference from the average value of B pictures (average value of Nb [0] to Nb [7]) is the maximum value. The average number AVR_Np of P pictures and the average number AVR_Nb of B pictures are obtained by removing GOPs including Nb [6] as minimum values. Further, GOPs may be arbitrarily selected and deleted one by one from GOPs having a plurality of maximum values and a plurality of minimum values without taking such a method.

  Further, the GOP excluded when predicting the GOP structure may be only one of the maximum value and the minimum value. Furthermore, a plurality of forward-predicted encoded images in a plurality of GOPs may be removed in descending order, or a plurality may be removed in ascending order. For example, in FIG. 11, both Np [2] and Np [3] may be removed, or both Np [6] and Np [7] may be removed.

  In the above-described embodiment, the moving image re-encoding device 100 has been described. However, a program that runs on a computer may be used. That is, the moving image re-encoding device 100 is replaced with a CPU, RAM, and ROM, and a moving image re-encoding program is recorded in the ROM, and the decoding unit 101, the encoding unit 103, and the encoded information analysis unit 102 are stored in the CPU. Have the function of. FIG. 12 shows a flowchart of the moving image re-encoding program.

  First, the input bit stream is decoded in step S401 as decoding means. Next, GOP structure prediction is performed in step S402 as image group structure prediction means. The content of step S402 is the same as that of the flowchart of FIG. Next, the encoding frame rate is predicted in step S403 as a frame rate prediction means. The content of step S403 is the same as that of the flowchart of FIG. Next, the allocated code amount per GOP is predicted in step S404 as the code amount predicting means. The content of step S404 is the same as that of the flowchart of FIG. Then, re-encoding is performed in step S405 as encoding means. By repeating the above steps for each picture, it is possible to function as a moving image re-encoding program.

  In the above-described embodiment, the picture is described as a frame, but the present invention can also be implemented in the field.

  The decoding means and the encoding means in the present invention are MPEG-1, 2, 4 and H.264. H.264 or other intra-coded image, forward prediction encoded image, and moving image that is a bit stream composed of an image group including at least one intra-coded image among bidirectional predictive encoded images Any combination may be used as long as it is an encoding method. Of course, you may use for the bit rate conversion between the same encoding systems.

  In addition, if the present invention is applied when broadcasting a moving image at a broadcasting station or transmitting it to a communication line, a moving image with a low delay and a low cost can be obtained with an encoding method or bit rate that matches the band of the broadcast wave or communication line. The image can be re-encoded. Further, for example, when the present invention is applied when recording an encoded moving image such as terrestrial digital broadcasting on a hard disk recorder, a DVD recorder, a Blu-ray recorder, an HD DVD recorder, or the like, it is more efficient than before re-encoding. Can be recorded with a low delay and a low cost.

  According to the embodiment described above, the following moving image re-encoding device, moving image re-encoding method, and moving image re-encoding program can be obtained.

(Supplementary note 1) A bit stream composed of a GOP including at least one I picture among I pictures, P pictures, and B pictures is input, the bit stream is decoded, and image data 113 and encoding information 112 are decoded. A decoding unit 101 for obtaining
In the moving image re-encoding device 100 having the encoding unit 103 that encodes the image data 113 again based on the image data 113 and the encoding information 112 decoded by the decoding unit 101,
Based on the encoding information 112, the number of P pictures in each of the plurality of GOPs is counted, and at least one of the GOPs having the maximum value or the minimum value of the number of P pictures in each of the plurality of counted GOPs And a GOP structure prediction unit 200 that outputs to a coding unit 103 as a GOP structure in which an average value of P pictures and B pictures in a plurality of GOPs is excluded. 100.

  According to the moving image re-encoding apparatus 100, it is possible to predict a more average GOP structure without being influenced by a GOP having an extremely large number of P pictures that appear temporarily. Further, since there is no need to hold an extra image for the GOP structure prediction, the delay time and the amount of memory are reduced.

(Supplementary note 2) A bit stream composed of a GOP including at least one I picture among an I picture, a P picture, and a B picture is input, the bit stream is decoded, and image data 113 and coding information 112 are decoded. In the moving image re-encoding method for encoding the image data 113 again based on the decoded image data 113 and the encoded information 112,
Based on the encoding information 112, the number of P pictures in each of the plurality of GOPs is counted, and at least one of the GOPs having the maximum value or the minimum value of the number of P pictures in each of the plurality of counted GOPs A moving picture re-encoding method, wherein an average value of a P picture and a B picture in a plurality of GOPs excluding one or more is used as a GOP structure that is predicted.

  According to this moving image re-encoding method, it is possible to predict a more average GOP structure without being influenced by a GOP having an extreme number of P pictures that appear temporarily. Further, since there is no need to hold an extra image for the GOP structure prediction, the delay time and the amount of memory are reduced.

(Supplementary Note 3) A bit stream composed of a GOP including at least one I picture among an I picture, a P picture, and a B picture is input and decoded, and the image data 113 and the encoding information 112 are decoded. Obtaining step S401;
In the moving image re-encoding program that causes the computer to function as step S405 that re-encodes the image data 113 based on the image data 113 and the encoding information 112 decoded in step S401,
Based on the encoding information 112, the number of P pictures in each of the plurality of GOPs is counted, and at least one of the GOPs having the maximum value or the minimum value of the number of P pictures in each of the plurality of counted GOPs A computer-readable moving image re-encoding that causes a computer to function as step S402 that outputs to step S405 as a GOP structure in which an average value of P pictures and B pictures in a plurality of GOPs excluding one or more is excluded Program.

  According to this moving image re-encoding program, it is possible to predict a more average GOP structure without being affected by a GOP having an extreme number of P pictures that appear temporarily. Further, since there is no need to hold an extra image for the GOP structure prediction, the delay time and the amount of memory are reduced.

  In addition, the Example mentioned above only showed the typical form of this invention, and this invention is not limited to an Example. That is, various modifications can be made without departing from the scope of the present invention.

Claims (10)

A bit stream composed of an image group including at least one of the intra-coded images among the intra-coded image, the forward predictive coded image, and the bidirectional predictive coded image is input to the bit stream. Decoding means for obtaining image data and encoded information;
In the moving image re-encoding device, the encoding unit that encodes the image data again based on the image data and the encoding information decoded by the decoding unit,
Based on the encoding information, the number of the forward prediction encoded images in each of the plurality of image groups is counted, and the maximum number of the forward prediction encoded images in each of the plurality of the image groups thus counted Or the average value of the forward prediction encoded image and the bidirectional predictive encoded image in a plurality of the image groups excluding at least one of the image groups of the image group having the minimum value, A moving image re-encoding apparatus comprising: an image group structure prediction unit that outputs the predicted structure of the image group to the encoding unit.   The image group structure prediction means excludes one or more of the image groups having the maximum value and the minimum value from the number of forward prediction encoded images in each of the counted plurality of images, respectively. The moving image re-encoding according to claim 1, wherein an average value of the forward prediction encoded image and the bidirectional predictive encoded image in a group is output to the encoding unit as a structure of the image group. apparatus.   The forward prediction in the plurality of image groups excluding the plurality of image groups in descending order from the number of the forward prediction encoded images in each of the plurality of counted image groups. The moving image re-encoding apparatus according to claim 1 or 2, wherein an average value of the encoded image and the bidirectional predictive encoded image is output to the encoding means as the structure of the image group.   The forward prediction code in the plurality of image groups excluding the plurality of image groups in ascending order of the number of forward prediction encoded images in each of the plurality of counted image groups. 4. The moving image re-encoding according to claim 1, wherein an average value of the encoded image and the bi-predictive encoded image is output to the encoding unit as the structure of the image group. 5. Device.   The image group structure prediction means counts the number of the bidirectional prediction encoded images in each of the plurality of image groups, and among the counted number of the forward prediction encoded images in each of the plurality of image groups counted, When there are a plurality of image groups each having a maximum value or a minimum value, an average of the bidirectional predictive encoded images in the plurality of image groups counted among the image groups having a plurality of maximum values or minimum values. An average value of the forward prediction encoded image and the bidirectional prediction encoded image in a plurality of the image groups excluding the image group having a value far from the value is output to the encoding unit as the structure of the image group The moving image re-encoding device according to any one of claims 1 to 4, wherein:   Based on the encoding information, the number of encoded images in each of the plurality of image groups and the number of images to be displayed in each of the plurality of image groups are counted, and the codes in each of the plurality of image groups are counted. Rate prediction for predicting a frame rate at which the encoding means encodes from the ratio between the number of converted images and the number of images to be displayed in each of the plurality of image groups and the encoding information The moving image re-encoding device according to any one of claims 1 to 5, further comprising: means.   Based on the structure of the image group predicted by the image group structure prediction unit, the frame rate predicted by the frame rate prediction unit, and the bit rate when the encoding unit is re-encoded in advance. The moving image re-encoding according to any one of claims 1 to 6, further comprising: a code amount predicting unit that predicts a code amount per group of images when the encoding unit performs encoding. Device. A bit stream composed of an image group including at least one of the intra-coded images among the intra-coded image, the forward predictive coded image, and the bidirectional predictive coded image is input to the bit stream. In the moving image re-encoding method, the image data and the encoding information are obtained, and the image data is encoded again based on the decoded image data and the encoding information.
Based on the encoding information, the number of the forward prediction encoded images in each of the plurality of image groups is counted, and the maximum number of the forward prediction encoded images in each of the plurality of the image groups thus counted Or the average value of the forward prediction encoded image and the bidirectional predictive encoded image in a plurality of the image groups excluding at least one of the image groups of the image group having the minimum value, A moving image re-encoding method, which is used at the time of encoding as the predicted structure of the image group. A bit stream composed of an image group including at least one of the intra-coded images among the intra-coded image, the forward predictive coded image, and the bidirectional predictive coded image is input to the bit stream. Decoding means for obtaining image data and encoded information;
In a moving image re-encoding program that causes a computer to function as an encoding unit that encodes the image data again based on the image data and the encoding information decoded by the decoding unit,
Based on the encoding information, the number of the forward prediction encoded images in each of the plurality of image groups is counted, and the maximum number of the forward prediction encoded images in each of the plurality of the image groups thus counted Or the average value of the forward prediction encoded image and the bidirectional predictive encoded image in a plurality of the image groups excluding at least one of the image groups of the image group having the minimum value, A computer-readable moving image re-encoding program that causes a computer to function as an image group structure prediction unit that outputs the predicted structure of the image group to the encoding unit. A computer-readable recording medium storing the moving image re-encoding program according to claim 9.

Images