JP2001197424A - Video editing apparatus, video editing method, and recording medium storing the program - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To provide a video editing apparatus capable of eliminating inconsistency of quantization matrix information and preventing image quality deterioration due to scene combination. SOLUTION: A quantization matrix used for decoding each scene is extracted based on scene information, and a scene is cut out based on the scene information. If the quantization matrix information of the scenes before and after the joining does not match, the quantization matrix information adjusting unit 104 adds the extracted quantization matrix information to each of the cut out scenes, combines the scenes, and creates a new stream. I do.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a video editing apparatus and related technologies, and more particularly, to an MPEG (Motion P).
The present invention relates to a technique of cutting out a plurality of continuous frames (scenes) from a stream coded based on the icutre Expert Group standard and creating a new stream by combining a plurality of scenes.

[0002]

2. Description of the Related Art MPE is an international standard for video coding.
In recent years, MPEG application fields are expanding, for example, MPEG1 is used for video CD and PC video data, and MPEG2 is used for DVD and digital satellite broadcasting.

The MPEG system is based on the International Standards Organization (ISO)
Is a moving picture coding system that defines the interpretation of a bit stream and its decoding process. There is ISO11172 as an MPEG1 standard, and IS as an MPEG2 standard.
O13818.

[0004] Among them, the MPEG1 standard is 1.5 Mb.
A compression method for compressing and storing video data in a digital recording medium having a transfer rate of about ps is defined.

[0005] The MPEG2 standard is an extension of the MPEG1 standard, and defines a compression method that takes into account not only storage media but also communication and broadcast media.

[0006] MPEG1 video data is composed of images having a frame structure, and each image is compressed using an intra-image correlation and an inter-image correlation. There are three types of image compression methods (image types): an intra-coded image (I picture), a forward predictive coded image (P picture), and a bidirectional predictive coded image (B picture).

[0007] The I picture has no correlation with other frames, and is coded using only the data of that frame. P
Picture is a temporally previous frame (past frame)
And is coded. The B picture is encoded by correlating both the temporally previous frame (past frame) and the subsequent frame (future frame).

FIG. 8 shows a correlation between images of MPEG1 data. In FIG. 8, a rectangle means one image, symbols inside the rectangle are symbols starting with I, I symbols are used, symbols starting with P are P pictures, and B symbols are symbols.
A symbol starting with "" means a B picture. In the following description, each image type is indicated according to this rule.

The images are arranged in the order of reproduction (from left to right in FIG. 8). In FIG. 8, arrows indicate the correlation between the images. It is encoded by referring to P4.

As described above, since there is a possibility of using the inter-image correlation with the subsequent image, the MPEG1 including the B picture is used.
As shown in FIG. 9, the order in which data is reproduced (reproduction order, upper row) differs from the order in which data is stored in the medium (accumulation order, lower row).

In general, the compression efficiency is (I picture) <(P picture) <(B picture), and the coding amount is conversely: (I picture)> (P picture)> (B picture) Becomes

Next, decoding processing for extracting a reproduced frame from an MPEG encoded stream will be described with reference to FIG. FIG. 10 is an explanatory diagram of a process of decoding MPEG video data. As shown in FIG. 10, the encoded data includes variable-length encoding, reverse scanning, inverse quantization, and inverse DCT.
(Discrete cosine transform), and through each step of motion compensation,
Decoded into decoded pixels.

In the step of performing inverse quantization, an effective quantization matrix is used for each image. This quantization matrix is set at the time of encoding,
If the quantization matrix used for decoding is incorrect,
The inverse quantization becomes incorrect, and as a result, the image quality of the decoded image is greatly deteriorated.

Also, MPEG1 video data is shown in FIG.
And a sequence header, a group of picture header (GOP header), a picture header, and picture data (slice data and thereafter) are described in this order.

[0015] The picture data is followed by a sequence header, a GOP header, or a picture header. Finally, a sequence end code indicating the end of the sequence is described. Further, extended data or user data can be entered.

Now, the quantization matrix in the stream is updated at the timing of the sequence header,
Thereafter, until the sequence header appears, the current quantization matrix is used. The color difference format used for the image of the MPEG1 video data is 4: 2: 0.
There are two types of quantization matrices, one for intra blocks and one for non-intra blocks. These two types of quantization matrices can be described or not described in the sequence header. In this case, a quantization matrix having a default value specified by the standard is used for decoding the subsequent image.

Also, MPEG2 has some differences from MPEG1, as shown in FIG. The first difference is that the sequence extension always follows the sequence header. Depending on the presence or absence of this sequence extension, MP
It is determined whether it is EG1 or MPEG2. That is, if there is no sequence extension, it is treated as MPEG1 video data, and if there is no sequence extension, it is treated as MPEG2.

The second difference is that the GOP header may not be provided, and the third difference is that the picture coding extension is always inserted after the picture header.

The quantization matrix is updated with a sequence header or a quantization matrix extension. The case of updating with the sequence header is the same as the case of MPEG1. The extension of the quantization matrix is shown in extension_and_user_data in FIG.
It can enter the position of (2). When the quantization matrix extension is included, the updated quantization matrix is used in subsequent images. In this case, only the type of quantization matrix described is updated, and for the type of quantization matrix not described, the previously updated quantization matrix continues to be used.

The color difference format of MPEG2 is
There is 4: 2: 2 or 4: 4: 4 as well as 4: 2: 0. In the case of 4: 2: 2 and 4: 4: 4, there are four types of quantization matrices. In the case of 4: 2: 0, two types of quantization matrices are used for the intra block and the non-intra block, for luminance and color difference, respectively. Divided into two.

Here, Japanese Patent Application Laid-Open No. Hei 10-13783 discloses an editing technique in which a plurality of continuous frames are cut out from MPEG video data as scenes, and a plurality of scenes are combined to create a new stream. ing.

Hereinafter, this prior art will be described with reference to FIG. First, an edit start image and an edit end image are input (step 1301), then an edit range is cut out (step 1302), and an edit range is output (step 1303). Then, it is determined whether there is any more edit target (step 1304).
The second and subsequent processes are repeated, and if not, the process is terminated.

Hereinafter, the clipping process in step 1302 will be described with reference to FIGS.

First, the processing of the cutout start portion will be described with reference to FIG. FIG. 14A shows the arrangement of frames in the reproduction order, and FIG. 14B shows the arrangement of frames in the accumulation order.

Here, when editing is started from the image I2 on the reproduced image shown in FIG.
As shown in the figure, the images B0 and B1 which may have been compressed by using the image I2 are deleted, and G is added before the image I2.
An OP header is added. When editing is started from the image P5 on the reproduced image shown in FIG.
As shown in FIG. 4 (d), the image P5
Is re-encoded into an image I5, and the images B3 and B4 which may have been compressed using the image P5 are deleted.
GOP header is added before 5. Further, FIG.
When editing is started from the image B3 on the reproduced image shown in (a), as shown in FIG.
5, images B3 and B4 are converted to images I3 and I3, respectively.
4 and rearranged in the order of reproduction.
An OP header is added.

Next, with reference to FIG. 15, the processing of the cutout end portion will be described. FIG. 15A shows the arrangement of frames in the order of accumulation, and FIG. 15B shows the arrangement of frames in the order of reproduction.

Here, when editing is to be ended with the image I2 on the reproduced image shown in FIG. 15B, the image may be compressed using the image I2 as shown in FIG. The remaining images B0 and B1 are left behind, and then a GOP header in the next editing range is added. When the editing is ended with the image P8 on the reproduced image shown in FIG.
As shown in FIG. 15D, images B6 and B7 which may be compressed using the image P8 are left, and a GOP header of the next editing range is added thereafter. further,
When the editing is ended with the image B6 on the reproduced image shown in FIG. 15B, as shown in FIG.
The image B6 is re-encoded into the image I6 by using the image P6 and the image P6, the image I6 is placed after the GOP header in the next editing range, and the image P8 is deleted.

[0028]

According to this conventional technique, scenes can be combined to create a new stream. However, in the related art, there is no consideration regarding the quantization matrix information before and after the combination. Therefore, when the quantization matrix information is different between the scenes before and after the combination, there is a problem that the inverse quantization is incorrect and the image quality is deteriorated.

It is an object of the present invention to provide a video editing apparatus capable of preventing image quality deterioration due to scene combination and its related technology.

[0030]

According to the present invention, the scene information input means inputs scene information by setting a plurality of continuous frames cut out from the stream to be edited as a scene. A quantization matrix extraction unit extracts a quantization matrix used for decoding each scene based on the scene information. A scene cutout unit cuts out a scene based on the scene information. The quantization matrix information adjusting means adds the extracted quantization matrix information to each cut-out scene when at least the quantization matrix information of the scenes before and after the combination does not match. Then, a scene combining unit combines the scenes to create a new stream.

With this configuration, even when scenes are combined, no inconsistency occurs in the quantization matrix information, so that image quality degradation can be prevented.

[0032]

According to the first, fifth and ninth aspects, a plurality of continuous frames cut out from a stream to be edited are set as scenes, scene information is input, and a plurality of continuous frames cut out from a stream to be edited are set as scenes. , Input scene information, extract the quantization matrix used for decoding each scene based on the scene information, and cut out the scene based on the scene information,
If at least the quantization matrix information of the scenes before and after the combination does not match, the extracted quantization matrix information is added to each cut-out scene, and the scenes are combined to create a new stream.

According to this configuration, even by scene combination,
To avoid inconsistency of the quantization matrix information,
Image deterioration can be prevented.

According to the second, sixth and tenth aspects, when the quantization matrix information before and after the combination is the same, the extracted quantization matrix information is not added to each of the cut out scenes.

With this configuration, useless information addition can be prevented.

According to the third, seventh and eleventh aspects, the extracted quantization matrix information is added without judging whether the quantization matrix information before and after the combination matches.

With this configuration, the processing speed can be improved by omitting the coincidence / mismatch determination of the quantization matrix information.

According to the fourth, eighth, and twelfth aspects, the quantization matrix information is information of a sequence header or a quantization matrix extension.

With this configuration, MPEG1, MPEG2
Can be handled for both.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of a video editing device according to an embodiment of the present invention.

In FIG. 1, scene information input means 101
, A plurality of continuous frames cut out from the stream to be edited are set as scenes, and scene information is input.

The quantization matrix extraction means 102 extracts quantization matrix information used for decoding each scene based on the scene information. The quantization matrix information may include a quantization matrix itself and a quantization matrix extension.

The scene cutout means 103 cuts out a scene based on the scene information.

The quantization matrix adjusting means 104 adds the extracted quantization matrix information to each of the cut-out scenes at least when the quantization matrix information of each scene before and after the combination does not match. by this,
It is possible to prevent inconsistencies in the quantization matrix due to scene combination.

However, if the quantization matrix information of each scene before and after the combination matches, the extracted quantization matrix information may not be added to each cut-out scene. In this way, useless information addition can be prevented, and the size of a new stream can be reduced, albeit slightly.

However, regardless of whether the quantization matrix information of each scene matches or not, in any case, the extracted quantization matrix information may be added to each cut-out scene. This makes it possible to omit the determination of the match / mismatch of the quantization matrix information, thereby improving the processing speed.

As will be described later, in the present embodiment, when the quantization matrix information of each scene before and after the combination matches, the extracted quantization matrix information is not added to each cut-out scene. However, the present invention is not limited to this.

The scene combining means 105 shown in FIG. 1 combines the scenes to create a new stream.

FIG. 2 is a hardware configuration diagram of the apparatus. The CPU 205 controls various operations of peripheral devices according to a control program stored in the external storage device 207 or the like. As peripheral devices, a display 201, a video RAM 202, a mouse 20
3, a keyboard 204, a main storage device 206, an external storage device 207, and a speaker 208 are connected by a bus 209.

The video RAM 202 stores the display 20
1 stores the image data to be displayed. Keyboard 204
When a key on the keyboard 204 is operated, the key code corresponding to the operation key is given to the CPU 205. When a button on the mouse 203 is operated, the mouse 203 gives a code corresponding to the button to the CPU 205, and when the mouse 203 moves, gives a code corresponding to the movement to the CPU 205. The units 101 to 105 shown in FIG. 1 are realized by the CPU 205 controlling these peripheral devices and executing a control program.

Now, the overall processing procedure of the present invention will be described with reference to FIG. First, the operator uses the scene information input unit 101 to input scene information by setting a plurality of continuous frames cut out from the stream to be edited as a scene (step 1). Next, the quantization matrix extraction unit 102 Based on the information, quantization matrix information used for decoding each scene is extracted (step 2).

Next, the scene cutout means 103 cuts out a scene based on the scene information (step 3).

Then, the quantization matrix adjusting means 10
4 examines whether or not the quantization matrix information of each scene before and after the combination matches (step 4). If they do not match, the extracted quantization matrix information is added to each cut-out scene (step 5). If they do not match, the addition of the quantization matrix information is omitted.

Then, the scene combining means 105 combines the scenes to create a new stream (step 6).

Next, the details of each step will be described. In step 1, scene information is input.

The scene information is information for specifying a plurality of continuous frames cut out from the stream to be edited. In the present embodiment, the scene information includes the number of scenes to be combined, the order of combination, cut-out image information, and an output file name.

The cut-out image information is information for designating a material to be combined and its section. The cut-out image information includes the file name of the material, the byte position of the cut-out leading image, and the byte position of the cut-out end image, and exists as many as the cut-out images to be combined. The output file name is an output destination file name when each scene is combined to create a new stream in step 6.

In step 2, the quantization matrix information used for decoding each scene is extracted based on the scene information. In the stream to be edited, the sequence header or the quantization matrix extension is searched backward from the top image of each scene, and the process ends when all the quantization matrices used in the stream to be edited have been extracted.

Here, the quantization matrix to be extracted is
When the chrominance format of the stream to be edited is 4: 2: 0, there are two types, one for the intra block and the other for the non-intra block, and when the chrominance format is 4: 2: 2 or 4: 4: 4, the intra There are four types: a luminance block, an intra chrominance block, a non-intra luminance block, and a non-intra chrominance block.

Next, a specific description will be given with reference to FIG. In FIG. 4, the color difference format indicated by the material 1 is 4: 2:
2 shows an example of the position of the sequence header and the quantization matrix extension in the stream to be edited which is 2.

Assuming that a backward search is performed from the first image of scene 1, first, the quantization matrix expansion Qma
t_ext (1) appears. If the quantization matrices described in the quantization matrix extension Qmat_ext (1) are the four types of quantization matrices, those quantization matrices are extracted and the processing is terminated. Otherwise, the quantization matrix described in the quantization matrix extension Qmat_ext (1) is extracted, and the backward search is continued.

Next, the sequence header Seq_H (2)
Appears. Among the quantization matrices described in the sequence header Seq_H (2), a type of quantization matrix other than the already extracted quantization matrix is extracted.

At the time when the sequence header Seq_H (2) appears, if there are any types of quantization matrices that have not been extracted yet, those matrices are extracted as default value quantization matrices.

This is because, after the appearance of the sequence header, even if there is no description of the quantization matrix in the sequence header, all of those types of quantization matrices are reset to default values.

In step 3, each scene is cut out based on the scene information. First, the minimum number of images (target images) that need to be re-encoded in order to enable each scene to be reproduced independently is specified. In the images included in the scene, the images other than the target image use the original image as it is. Here, in this specification, a range in which the original image can be used as it is is referred to as a “reuse range”.

Next, the image type of the target image is determined.
Finally, re-encoding is performed using the determined image type for the target image. With the above processing, each scene can be independently reproduced.

Further, regarding the processing of step 3, FIG.
This will be specifically described with reference to FIG. In FIG. 5, the images of the stream to be edited are arranged in the order in which they are reproduced.

Now, it is assumed that a scene including the images B3 to B11 is designated, and this range is cut out. At this time, image B
3, P4, and B11 refer to images outside the range specified in the scene, so there is no reference source image when cutting out, and it is impossible to reproduce alone.

Accordingly, these images B3, P4 and B1
1 as target images, and images I3 ′, I4 ′ and I
11 'is re-encoded. At this time, the quantization matrix used for the re-encoding uses the quantization matrix used for the target image. As a result, the reference source of the image included in the range specified by the scene is closed within the scene, and the scene can be reproduced independently.

In FIG. 5, the picture type of the target picture after re-encoding is I picture, but of course P, B
Re-encoding may be performed including a picture.

In step 4, it is examined whether or not the quantization matrix information of the scene before and after the combination matches,
If they do not match, in step 3, information of the quantization matrix extracted in step 2 is added to each cut-out scene (step 5).

Further, regarding the processing of step 4, FIG.
This will be specifically described with reference to FIG. FIG. 6 schematically shows the relationship between scene 1 and a quantization matrix that can be added to scene 1.

As shown in FIG. 6, by step 3,
Assuming that scene 1 is cut out, sequence header Seq_H (1) as quantization matrix information is added to the beginning of scene 1 as necessary. In this sequence header Seq_H (1), the quantization matrix used in the first image of scene 1 extracted in step 2 is described.

By doing so, it is possible to compensate for no inconsistency in the quantization matrix information. Therefore, when decoding the scene 1, the correct quantization matrix is used, and the image quality does not deteriorate. In addition, since quantization matrix information can also be performed by using quantization matrix extension, instead of adding a sequence header, when re-encoding the first image, the quantization matrix that describes the extracted quantization matrix is used. A matrix extension can also be inserted.

In step 5, each scene is combined to create a new stream. Based on the scene information input in step 1, each scene is combined in a specified order, and a new stream having an output file name specified by the scene information is created.

FIG. 7 shows a specific example. If it is specified in step 1 that three scenes are to be combined in the order of scenes 1, 2, and 3, and if the quantization matrix information of each of scenes 1, 2, and 3 is all different, step 4
By the processing of (1), the target image is re-encoded and cut out so that each scene can be reproduced independently.
As shown in (a), at the beginning of each scene, quantization matrix information (see hatched portions; sequence header or quantization matrix extension) describing a correct quantization matrix is added.

At this time, in step 6, as shown in FIG. 7B, the scenes are combined in the specified order. First, the sequence headers at the beginning of scene 1 are combined, and the re-encoded target images at the beginning of scene 1 are combined in the order of accumulation. At this time, if the image type of the target image before editing is different from the image type after re-encoding, the target image is rearranged in the accumulation order according to the image type after re-encoding of the target image based on the reproduction order. Next, the reuse ranges of scene 1 are combined.

At the end of the scene 1, the re-encoded image at the end is combined in the order of accumulation. The same applies to scene 2 and scene 3. The new stream created by combining is output to a file with the specified output file name. In step S6, file output is performed after scenes are combined, but file output can also be performed while sequentially combining scenes.

Further, assuming that the quantization matrix information matches between scene 1 and scene 2 and the quantization matrix information does not match between scene 2 and scene 3, it is assumed that the quantization matrix information Since the addition can be omitted, as shown in FIG. 7C, the stream can be a stream that is slightly smaller in size than the stream in FIG. 7B.

With the above processing, a plurality of continuous frames (scenes) are cut out from the stream to be edited, and a new stream in which a plurality of scenes are combined is created.

Note that a plurality of scenes may be designated for one stream, and the range of each scene may overlap. Of course, a scene may be designated for a plurality of streams.

In this embodiment, the editing process of the coded stream is realized by a program. However, a part or the whole of this process may be configured by hardware such as a logic circuit. With this configuration, the processing time can be reduced.

[0083]

According to the present invention, a plurality of continuous frames (scenes) are cut out from an existing MPEG stream,
By combining a plurality of scenes, it is possible to create a new stream without image quality deterioration due to mismatch of quantization matrices used for decoding.

[Brief description of the drawings]

FIG. 1 is a block diagram of a video editing apparatus according to an embodiment of the present invention.

FIG. 2 is a hardware configuration diagram of the apparatus.

FIG. 3 is a flowchart showing an overall processing procedure of the apparatus.

FIG. 4 is an explanatory diagram of a quantization matrix extraction procedure;

FIG. 5 is an explanatory diagram of cutout processing of the scene.

FIG. 6 is an explanatory diagram showing a relationship between the scene and quantization matrix information to be added;

7A is a diagram illustrating the relationship between each scene and quantization matrix information. FIG. 7B is a diagram illustrating the structure of the stream after the combination (no omission). FIG. 7C is a diagram illustrating the structure of the stream after the combination.

FIG. 8 is an explanatory diagram of a conventional scene combining process.

FIG. 9 is a diagram showing the relationship between the reproduction order and the storage order of MPEG data images.

FIG. 10 is an explanatory diagram of a decoding process.

FIG. 11 is a structural diagram of MPEG1 video data.

FIG. 12 is a structural diagram of MPEG2 video data.

FIG. 13 is a flowchart of a conventional video editing device.

FIG. 14 is an explanatory diagram of a conventional clipping process (starting part).

FIG. 15 is an explanatory diagram of a conventional clipping process (end section).

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 101 Scene information input means 102 Quantization matrix information extraction means 103 Scene extraction means 104 Quantization matrix information adjustment means 105 Scene connection means

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Atsushi Ikeda 1006 Kadoma Kadoma, Kadoma, Osaka Prefecture Inside Matsushita Electric Industrial Co., Ltd. (72) Yuichi Kawaguchi 1006 Kadoma Kadoma, Kadoma City, Osaka Matsushita Electric Industrial F Term (reference) 5C053 FA14 GB06 GB08 GB09 GB22 GB26 GB34 GB37 GB38 GB40 JA21 JA30 KA01 LA06 5C059 KK01 KK39 MA00 MA23 MC14 ME01 PP05 PP06 PP07 PP16 RB09 RC14 RC26 RC31 SS20 UA02 UA05 UA29 UA39 5L096 EA24 EA24 EA24 EA24 EA24 EA24 EA24 EA24 EA24 EA24 EB

Claims (12)

[Claims] 1. A scene information input means for inputting scene information by using a plurality of continuous frames cut out from a stream to be edited, and a quantization matrix used for decoding each scene based on the scene information. Extracting quantization matrix extracting means, scene extracting means for extracting a scene based on the scene information, and extracting the quantized matrix into each extracted scene when at least the quantization matrix information of the scene before and after the combination does not match. A video editing apparatus comprising: quantization matrix information adjusting means for adding quantization matrix information; and scene combining means for combining scenes to create a new stream. 2. The quantization matrix information adjusting means according to claim 1,
When the quantization matrix information before and after combining matches,
2. The video editing apparatus according to claim 1, wherein the extracted quantization matrix information is not added to each of the extracted scenes. 3. The quantization matrix information adjusting means,
2. The video editing apparatus according to claim 1, wherein the extracted quantization matrix information is added to each of the cut-out scenes without determining whether the quantization matrix information before and after the combination is equal. 4. The video editing apparatus according to claim 1, wherein the quantization matrix information is information of a sequence header or a description of extension of the quantization matrix. 5. A scene information input step of inputting scene information using a plurality of continuous frames cut out from a stream to be edited, and a quantization matrix used for decoding each scene based on the scene information. Extracting a quantization matrix, extracting a scene based on the scene information, and extracting a scene.If at least the quantization matrix information of the scene before and after the combination does not match, the extracted quantum is added to each of the extracted scenes. A video editing method, comprising: a quantization matrix information adjusting step of adding quantization matrix information; and a scene combining step of combining respective scenes to create a new stream. 6. In the quantization matrix information adjusting step, when the quantization matrix information before and after the combination matches, the extracted quantization matrix information is not added to each cut-out scene. Video editing method described. 7. The quantization matrix information adjusting step is characterized in that extracted quantization matrix information is added to each cut-out scene without judging whether the quantization matrix information before and after the combination is coincident. The video editing method according to claim 1. 8. The video editing method according to claim 1, wherein the quantization matrix information is information of a sequence header or an extension description of the quantization matrix. 9. A scene information input step of inputting scene information using a plurality of continuous frames cut out from a stream to be edited, and a quantization matrix used for decoding each scene based on the scene information. Extracting a quantization matrix, extracting a scene based on the scene information, and extracting a scene.If at least the quantization matrix information of the scene before and after the combination does not match, the extracted quantum is added to each of the extracted scenes. A recording medium recording a video editing program, comprising: a quantization matrix information adjusting step of adding quantization matrix information; and a scene combining step of combining respective scenes to create a new stream. 10. In the quantization matrix information adjusting step, when the quantization matrix information before and after the combination matches, the extracted quantization matrix information is not added to each of the cut out scenes. A recording medium on which the described video editing program is recorded. 11. The quantization matrix information adjusting step is characterized in that the extracted quantization matrix information is added to each of the cut out scenes without judging whether the quantization matrix information before and after the combination matches. A recording medium on which the video editing program according to claim 1 is recorded. 12. The recording medium according to claim 1, wherein the quantization matrix information is information of a sequence header or a quantization matrix extension description.