JP2000076462A - Moving image scene detector - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a moving image scene detector capable of decoding only attribute information in encoded/recorded moving image information and detecting the candidate of a characteristic scene through the use of this attribute information. SOLUTION: Attribute information in code information stored in an information storing means 101 is decoded by an attribute information decoding means 102, the candidate of the characteristic scene in the moving image is detected by a candidate scene detecting means 103 through the use of this decoded attribute information, image information in the code information is decoded by an image information decoding means 104 and this decoded image information is displayed by an image information display means 105.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for detecting and recording a characteristic scene from moving image information compressed and recorded on a digital recording medium.

[0002]

2. Description of the Related Art For example, when editing a moving image in a movie or television, a desired portion is cut out from a material film or a material video tape and connected.

[0003] Basic editing of a moving image mainly includes the following three procedures.

1) A desired scene is searched from a plurality of pieces of video information.

2) Cut out the searched scene.

3) Connect the extracted scene to a desired sequence.

[0007] Of these three, the most problematic in the efficiency of editing work is the scene search 1). The reasons are as follows.

[0008] Conventionally, a scene search is performed visually, which takes time and effort. Scene search requires the skill of the editor.

On the other hand, in recent years, the technology for encoding digital moving pictures has been advanced, and MPEG (Motion Picture E), which is being studied as an international standard for storage-based moving picture encoding, has been studied.
xpert Group), a moving image for about one hour can be converted into one CD-R.
It has become possible to memorize in OM. In such a situation, the demand for handling a large amount of video information is increasing in combination with the economics of the storage medium, and the efficiency of the 1) scene search is considered to be a major problem.

Against this background, techniques for automatically performing scene retrieval have been studied.

For example, Ueda: "Proposal of Interactive Video Editing System", IEICE Technical Report, Vol. IE90-
No. 6,1990 discloses an outline of a technique for performing a scene search using a feature amount of an image.

Also, Nagasaka and Tanaka: "Automatic detection of scene change in video work", Proc. Of the 40th Annual Meeting of the Information Processing Society of Japan (Early 1990), 1Q-5, p. 642-
643, 1990 discloses a method of detecting a place where inter-frame correlation becomes particularly low as a scene change.

Japanese Patent Application Laid-Open No. 02-404272 proposes a device that supports scene change detection based on the size of difference information from a frame. As shown in FIG. 13, this apparatus comprises a scene change determination unit 801, a moving image data unit 802, an attribute information decoding unit 803, a data decompression unit 804, and a display unit 805. Such attribute information including a moving image information body and an information amount, a frame position, and the like are recorded in frame units. For example, in the moving image data unit 802,
If the information obtained by encoding the difference information from the previous frame as the moving image information is recorded with the code amount and the frame position for each frame as the attribute information, scene change is performed by performing threshold processing on the code amount. Can be detected. That is, a frame having a large code amount (the amount of difference information from the previous frame) is a frame that has greatly changed from the previous frame, and thus it is considered that there is a high possibility that a scene change has occurred. However, even when the area of the moving object is large, the amount of difference information from the previous frame is large, so that a scene change is not necessarily detected by the code amount threshold processing.

Haseyama, Tanaka, and Oba: "Study of Moving Image Retrieval Using Encoded Information Amount", Proceedings of the 1992 IEICE Spring Conference, D-292, show that interframe predictive DCT coding is used. A technique for performing a scene search on an encoded moving image based on a code amount for each frame is disclosed.

[0015]

When editing or scene detection is performed by visual confirmation, it is necessary to reproduce (or fast-forward) the entire moving image information. For this reason, there are problems in efficiency, such as a long working time and difficulty in performing the work simultaneously using a plurality of pieces of moving image information. In addition, when moving image information is stored in a non-encoded state, the capacity of the storage device becomes a problem. On the other hand, when the moving image information is encoded and stored, the entire target moving image information must be decoded for detection. However, the decoding processing time is a problem.

Ueda: "Proposal of Interactive Video Editing System", IEICE Technical Report, Vol. IE90-6,1
990, and Nagasaka and Tanaka: "Automatic Detection of Scene Change in Video Works", IPSJ 40th (Heisei 2
1st semester), 1Q-5, pp. 642
In the method disclosed in US Pat. No. 6,643,1990, since a moving image that is not coded is targeted, there is a problem that the processing system requires a high processing capability and a large storage capacity.

In the apparatus disclosed in Japanese Patent Application Laid-Open No. 02-404272, a scene change determination unit 80 is provided.
Since the determination of Data> Th is performed only by using the information amount (Data) of the difference information between frames and the threshold value (Th) in 1, only one type of scene change can be detected. Further, since the recording of the detected scene is not performed, the user needs to perform the scene detection processing by the scene change detection support device every time the editing operation is performed.

Further, as described in the method of Haseyama, Tanaka, and Ohba, "Examination of Moving Image Retrieval Using Encoded Information Amount", Proc. Of the 1992 IEICE Spring Conference, D-292. In a case where a scene search is simply performed based only on the code amount for each frame, in a coding method in which the code amount is controlled so that the generated information amount is constant,
Highly accurate scene detection cannot be performed.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems. It is an object of the present invention to decode only attribute information of encoded and recorded moving image information. It is an object of the present invention to provide a moving image scene detection device capable of detecting a characteristic scene candidate using attribute information.

It is another object of the present invention to provide a moving image scene detecting device capable of performing highly accurate scene change detection based on various statistics obtained from the attribute information.

[0021]

In order to achieve the above object, a moving image scene detecting apparatus according to the present invention has a moving image information body which is encoded moving image information and an attribute indicating an attribute of the moving image information body. Information storage means for storing code information comprising information; attribute information decoding means for decoding attribute information in the code information; and characteristic information in a moving image using attribute information output from the attribute information decoding means. Candidate image detecting means for detecting a suitable scene candidate, image information decoding means for decoding image information in the code information, and image information display means for displaying image information output from the image information decoding means. It is characterized by having.

[0022]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be specifically described with reference to examples.

As shown in FIG. 1, the moving picture scene detecting apparatus according to the present invention comprises an information storage means 101 for storing code information composed of coded moving picture information and attribute information and editing information;
Attribute information decoding means 102 for decoding attribute information in code information stored in the information storage means 101, and characteristic scene candidates in a moving image using attribute information output from the attribute information decoding means 102 Candidate image detecting means 103 for detecting image information, and image information decoding means 1 for decoding image information in the code information stored in the code information storage means 101
04, and image information display means 105 for displaying decoded image information output from the image information decoding means 104.

In the above configuration, the information storage means 101
, Only the attribute information for each frame is read out, decoded by the attribute information decoding means 102, and a characteristic scene candidate (hereinafter referred to as a candidate) in a moving image using the decoded attribute information output from the attribute information decoding means 102. (Abbreviated as scene)
Is detected by the candidate scene detecting means 103. Every time a candidate scene is detected, the detected candidate scene and image information of several frames before and after the detected candidate scene are sequentially read out from the information storage means 101 and decoded by an image information decoding means 104, and the decoded image information is displayed in an image information display. They are displayed side by side on the means 105. The user looks at the image information displayed side by side on the image information display means 105, judges the type of the detected scene, and sets the type of the scene and the frame position as edit information of the code information in the information storage means 101. To record. At this time, the scene type is determined by the user and is not limited to one type. By performing the above operation on the entire moving image information, the editing information that is the type and position information of the characteristic scene in the moving image is stored in the information storage unit 101 together with the moving image information. Therefore, when the user performs editing work or the like, the editing information can be used without performing scene detection again, and efficiency can be improved.

[0025]

FIG. 2 shows the configuration of an embodiment of a moving picture scene detecting apparatus according to the present invention, FIG. 3 shows the configuration of code information in the present embodiment, and FIG. 3 shows the flow of signals.

In FIG. 2, 1 is an instruction input / output unit for inputting an instruction by a user, 2 is an information storage unit, 3 is an attribute information decoding unit, 4 is a candidate scene detection unit, 5 is an image information decoding unit, and 6 is an image information decoding unit. It is an image information display section. In FIG. 5, reference numeral 7 denotes a detection recording start instruction signal for instructing the user to start detection recording, 8 denotes attribute information read from the information storage unit 2, 9 denotes decoding attribute information output from the attribute information decoding unit 3, Reference numeral 10 denotes an attribute information reading instruction signal for instructing reading of attribute information from the information storage unit 2, reference numeral 11 denotes a designated frame image display instruction signal for instructing display processing of image information at a frame position detected by the candidate scene detecting unit 4, and reference numeral 12 denotes Image information designated by the designated frame image display instruction signal 11 and read from the information storage unit 2, 13 is decoded image information output from the image information decoding unit 5, and 14 is a command to record editing information and restart reading attribute information. 15 is a last frame instruction signal indicating that the detected candidate scene is the last frame, and 16 is recording and detection of editing information. A recording detection end instruction signal for instructing the end of the processing, a detection end instruction signal for instructing only the end of the detection processing, a frame position code indicating a frame position, and an inter-frame coding / intra-frame coding. , 20 is an information amount code indicating the code amount of the frame, and 21 is a vector number code indicating the number of non-zero motion compensation vectors in the frame.

The code information shown in FIG. 3 is obtained as a result of encoding the moving image information using the motion-compensated inter-frame / intra-frame adaptive prediction encoding method. The inter-frame / intra-frame adaptive predictive encoding method is an adaptive method between an inter-frame encoding method for encoding difference information from a previous frame and an intra-frame encoding method for encoding information of an encoding target frame as it is. This is a technique used by switching. In this method, the intra-frame coding method is applied to a frame having a large change from the previous frame, that is, a frame having a large amount of difference information. In addition, from the viewpoint of reducing image quality deterioration due to accumulation of coding errors and convenience in decoding, periodic refresh forcibly applying an intra-frame coding method is performed at a fixed cycle. FIG. 4 shows the principle of the motion compensation coding method. In this method, when calculating the difference from the previous frame, the motion of the object is taken into account, that is, a motion vector is detected, and the position is shifted to minimize the difference from the previous frame. Therefore, the magnitude of the motion vector is zero in the background portion and the area of the stationary object, and the larger the area of the moving object, the larger the number of non-zero motion vectors in the frame. In particular, when panning, zooming, or the like has occurred, the motion vector is considered to be non-zero in most areas.

Next, the operation will be described. In FIG. 5, when the user inputs a detection recording start instruction and an image information name from the instruction input / output unit 1, a detection recording start instruction signal 7 is sent to the information storage unit 2, and the first frame of the designated image information is transmitted. Is read from the information storage unit 2 and sent to the attribute information decoding unit 3 as attribute information 8, and the attribute information decoded by the attribute information decoding unit 3 is sent to the candidate scene detection unit 4 as decoded attribute information 9. . Candidate scene detection unit 4
FIG. 6 shows the flow of the processing in.

First, upon receiving the decoding attribute information 9, the candidate scene detecting section 4 checks the decoding result of the interframe / inner code 19. In this embodiment, the inter-frame / inner code 19
Is defined as Inter_Intra, and
If _Intra = 0, interframe coding is performed, and Int
When er_Intra = 1, it indicates that encoding has been performed by intra-frame encoding. Inter_I
Since ntra = 1 indicates that the frame is a large change from the previous frame unless the frame is a periodic refresh frame, it is highly likely that the scene is a characteristic scene such as a scene change or the appearance of an object. Therefore, the designated frame image display instruction signal 11 based on the decoding result of the frame position code is sent to the information storage unit 2. Inter_Int
If ra = 0 and if the frame is a periodic refresh, the following determination processing is performed (step 601).

Next, the result of decoding the information amount code 20 in the attribute information 9 (this is referred to as mount) is examined. am
between the threshold value and the preset threshold value Th1
If the relationship t ≧ Th1 holds, it indicates that the frame has a large change from the previous frame to some extent, and thus it is highly possible that the scene is a characteristic scene such as a scene change or the appearance of an object. Therefore, the designated frame image display instruction signal 11 based on the decoding result of the frame position code is sent to the information storage unit 2. If amount <Th1, the next determination process is performed (step 602).

Thereafter, the decoding result of the vector number code 21 in the attribute information 9 (this is referred to as Number) is examined.
Num between Number and a preset threshold Th2
If the relationship of ber ≧ Th2 holds, it is highly possible that the scene is a characteristic scene such as panning. Therefore, the designated frame image display instruction signal 11 based on the decoding result of the frame position code is sent to the information storage unit 2. Number <Th
If it is 2, the next processing is performed (step 603).

Finally, it is checked whether or not the frame being processed is the last frame. If it is the last frame, the last frame instruction signal 15 is sent to the instruction input / output unit 1. If it is not the last frame, a frame attribute information read instruction signal 10 is sent to the information storage unit 2 (step 604).

In FIG. 5, when the attribute information read instruction signal 10 is sent to the information storage unit 2, the same processing as described above is performed on the frame next to the frame from which the attribute information was last read, so that the attribute information 8 is The attribute information is sent to the attribute information decoding unit 3, and the attribute information decoding unit 3 sends the decoded attribute information 9 to the candidate scene detection unit 4 to continue the scene detection.

When the designated frame image display instruction signal 11 is sent from the candidate scene detecting section 4 to the information storage section 2, the designated frame and several frames before and after the designated frame (the number of frames at this time are set in advance) are sequentially stored in the image information. The image information is transmitted to the image information decoding unit 5 as 12, sequentially decoded by the image information decoding unit 5, transmitted to the image information display unit 6 as decoded image information 13, and displayed side by side with time. However, since only the difference information from the previous frame is encoded in the frame encoded by the inter-frame encoding method, it is necessary to sequentially decode the frames encoded by the intra-frame encoding method as a starting point. No reproduced image can be obtained.

FIG. 7 shows a display example on the image display unit 6. 7, reference numeral 701 denotes a detected candidate scene;
02, the designated number of images temporally preceding the candidate scene 701; 703, the designated number of images temporally behind the candidate scene 701;
Indicates a frame located two places before the detected candidate scene 701 among frames encoded by the intra-frame encoding method.

The user determines the type of the scene from the images of several frames centered on the candidate scene shown on the image display section 6. At this time, if the displayed scene has not reached the last frame, the recording input / output unit 1 instructs the recording of the scene type and the frame position and the restart of the detection process, so that the recording detection restart instruction signal 14 After being sent to the storage unit 2 and recording the editing information, the detection process is restarted. When the displayed scene has reached the last frame, the recording / editing end instruction signal 16 is sent to the information storage unit 2 by instructing the recording and editing of the editing information from the instruction input / output unit 1 to close the file. Is performed.

When the last frame instruction signal 15 is sent to the instruction input / output unit 1 in step 604, a message prompting the user to input a processing end instruction is displayed, and when the user instructs the end of the detection processing. , A detection end instruction signal 17 is sent to the information storage unit 2, and post-processing such as file closing is performed.

As a result of the detection / determination, the editing information recorded in the information storage unit 2 is displayed so that the editing information can be visually understood as shown in FIG. It is possible to make it easy for the user to grasp the position and type of a scene. FIG. 8 shows an example in which only information relating to a characteristic scene in moving image information is displayed.
Scene change in seconds, panning in 30 seconds, 1 minute 2
At 0 seconds, a scene change has occurred.

Next, another embodiment of the moving picture scene detecting apparatus according to the present invention will be described.

First, a moving picture encoder for generating a coded picture in which a scene change is detected by the moving picture scene detecting apparatus of the present invention will be described with reference to FIG. The moving picture encoder shown in FIG. 9 employs a coding scheme that combines “inter-frame prediction” and “DCT (discrete cosine transform)”.

In the block extracting circuit 30, an image block of the current frame consisting of a fixed number of pixels is extracted from the input image (current frame image) and supplied to the subtracter 31.
In the subtracter 31, the previous frame image block is subtracted from the current frame image block, and the difference image block is supplied to the quantization circuit 33 via the discrete cosine transform (DCT) circuit 32, and the quantization index (hereinafter, referred to as “transform coefficient”) (Referred to as conversion coefficient information). This transform coefficient information is supplied to the variable length coding unit 34 and is subjected to variable length coding. The transform coefficient information from the quantization circuit 33 is supplied to the inverse discrete cosine transform (IDC) via the inverse quantization circuit 35.
T) The difference image block supplied to the circuit 36 is reproduced. The difference image block is supplied to the frame memory 38 via the adder 37, and the frame memory 38
, Ie, the previous frame image block, is supplied to the subtractor 31 and the adder 37. The configuration up to this point is the same as the configuration of a general video encoder.

In the moving picture encoder shown in FIG. 9, the difference mode block from the subtractor 31 is stored in the coding mode control unit 40, and the transform coefficient from the quantization circuit 33 is stored in the significant / insignificant block control unit 39. Information is provided.

The coding mode control unit 40 determines that intra-frame coding should be applied (intra-frame coding mode) since inter-frame prediction does not function effectively for a block having a particularly large difference. It is determined that the inter-frame prediction should be applied to other blocks (inter-frame prediction mode), and the result of this determination is used as encoding mode information as the first switch circuit 41, the second switch circuit 42, and the like. Well, it is supplied to the variable length coding unit 34.

When the encoding mode information from the encoding mode control unit 40 is the intra-frame encoding mode, the first
In the switch circuit 41, the switch is connected to the terminal 41a, and the DCT circuit 32 is supplied with the current frame image block. When the encoding mode information is the inter-frame prediction mode, the switch in the first switch circuit 41 is connected to the terminal 41b, and the differential image block is supplied to the DCT circuit 32.

When the coding mode information from the coding mode control section 40 is the intra-frame coding mode, the switch in the second switch circuit 42 is connected to the terminal 42.
a, and the adder 37 is supplied with a "zero block" in which the values in the block are all zero. When the encoding mode information is the inter-frame prediction mode, the switch in the second switch circuit 42 is connected to the terminal 42b,
The output (previous frame image block) from the frame memory 38 is supplied to the adder 37.

When the coding mode information from the coding mode control section 40 is the intra-frame coding mode, the significant / insignificant block control section 39 controls the block where the transform coefficient information from the quantization circuit 33 becomes substantially zero. Is determined as an insignificant block, and the other blocks are determined as significant blocks, and the determination result is supplied to the variable length coding unit 34 as significant / insignificant block information.

The variable length coding unit 34 includes significant / insignificant block information from the significant / insignificant block control unit 39 and
The coding mode information from the coding mode control unit 40 and the transform coefficient information from the quantization circuit 33 are subjected to variable-length coding to generate a coded image, and the coded image is transmitted to the image decoding device via a transmission line. Or store it in a storage device. However, when the significant / insignificant block information from the significant / insignificant block control unit 39 is an insignificant block, the transform coefficient information from the quantization circuit 33 is not subjected to variable length coding. If the block is an insignificant block, it is supplemented by the block of the previous frame at the time of decoding. Also, here, the significant / insignificant block information and the encoding mode information are the attribute information in the encoded moving image information.

Next, the principle of detecting a scene change using the attribute information used in the above-described moving picture encoder will be described.

As described below, the characteristics of an image can be estimated to some extent from attribute information. As is clear from the above description, a significant block is a block that has changed from the previous frame. Therefore, a significant block is usually a moving area in a frame. However, a frame in which the number of significant blocks in one frame is particularly large is not a mere moving area, and a possibility that a scene change has occurred is high.

Also, the intra-frame coding mode is selected for a block in which significant changes from the previous frame are particularly severe even among significant blocks. Therefore, there is a high possibility that a scene change has occurred in a frame in which the ratio of intra-frame coding application blocks in a significant block is particularly large.

Therefore, in the present invention, a scene change is detected by focusing on the correlation between the attribute information and the scene change.

FIG. 10 is a block diagram showing an embodiment of the moving picture scene detecting apparatus according to the present invention.

The variable-length coded image supplied via the transmission line or read from the storage device is decoded by the variable-length decoding unit 51 and converted coefficient information and attribute information (significant information). / Insignificant block information, coding mode information). These transform coefficient information and attribute information are supplied to the decoding unit 52 to obtain a decoded image. The attribute information is also supplied to the scene change detection filter 53. This attribute information is the significant / insignificant block information and the encoding mode information itself described in the video encoder shown in FIG.

The scene change detection filter 53 includes two statistic calculation circuits 54a and 54b to which attribute information is supplied.
And the output of the statistic calculation circuits 54a and 54b
Comparators 55a and 55b for comparing with H ₁ and TH _2, and an AND gate 56 for taking the logical product of the outputs of the comparators 55a and 55b
It is composed of

The statistic calculation circuit 54 calculates the number of significant blocks in the frame as the statistic, and calculates the statistic calculation circuit 55.
Calculates the ratio of the intra-frame coding mode in the significant block as a statistic.

The characteristics of the scene change detection filter 53 are represented by the following equations.

[0057]

(Equation 1) D: scene change detection result d ₁ : determination result regarding statistic f _i f _i : i-th statistic function c (n): block characteristic information of n-th frame TH _i : threshold value for i-th statistic Accordingly, the scene change detection filter 53 shown in FIG. 10, the number of significant blocks in a frame is greater than the threshold value TH _1, and, when the ratio of the intraframe coding mode in significant blocks is greater than the threshold value TH ₂ Outputs a detection result indicating that a scene change has been detected.

In the embodiment shown in FIG. 10, a scene change detection result is obtained from two statistics, that is, the number of significant blocks in a frame and the ratio of the intra-frame coding mode in the significant block. Scene change detection can be performed.

Next, still another embodiment of the moving picture scene detecting apparatus according to the present invention will be described with reference to FIG. Parts corresponding to those in the embodiment shown in FIG. 10 are denoted by the same reference numerals. In the embodiment shown in FIG. 11, in addition to the number of significant blocks in the frame and the ratio of the intra-frame coding mode in the significant block, the code amount of the frame is employed as the statistic. Note that the code amount of a frame means the total amount of one frame of the code amount in block units output from the variable length coding unit. This code amount information is obtained by examining the code length of one frame of the image subjected to the variable length coding.

The statistic calculation circuit 54c of the scene change detection filter 53 obtains the code amount for one frame from the coded image. This code amount is compared with the threshold value TH _{3 by the} comparator 55c.
Is compared with the threshold value TH _3, and d ₃ = 1
Is output. The output d ₃ of the comparator 55c, the other two comparators 55a, 55b with the output d _1, d ₂ of, because it is supplied to the AND gate 56, the scene change detection result D is D = d ₁ · d It is expressed by ₂ · d ₃ .

As described above, in the embodiment shown in FIG. 11, the scene change detection is performed based on the three statistics of the number of significant blocks in the frame, the ratio of the intra-frame coding mode in the significant block, and the code amount of the frame. Since the result is obtained, it is possible to detect a scene change without errors.

An experiment was conducted to confirm the effect of scene change detection based on the above three statistics. Table 1 shows the experimental conditions.

[0063]

[Table 1]

FIG. 12 shows the experimental results. FIG. 7A shows the difference power between frames of the image to be tested, and FIGS.
(C) and (d) show the determination results of the code amount, the determination result of the number of significant blocks, and the individual determination results d ₃ , d ₁ , and d _{2 for} the respective statistics of the ratio of the intra-frame coding mode. (E) is the output D of the detection filter.

In the image sequence for the experiment, the difference power is increased in response to the scene change. In order to prevent omission of scene change detection, each comparator 55
the threshold TH _i of a~55c was set lower. For this reason, an individual detection result includes an erroneous detection. But 3
By taking the logical product of the two determination results, there is no erroneous detection in the output of the detection filter 53.

As can be seen from FIG. 12, by providing a scene change detection filter using the statistic obtained from the attribute information and the code amount of the frame, the scene change of the image sequence to be tested can be accurately detected.

This eliminates the need for visual confirmation, which was conventionally required, and can significantly improve the efficiency of moving image editing.

In the embodiments shown in FIGS. 10 and 11, similarly to the embodiment shown in FIG. 2, the result of the scene change detection may be recorded in the information storage together with the encoded moving image.

[0069]

As described above, according to the present invention, it is possible to decode attribute information in coded video information and detect characteristic scene candidates in the video based on the decoded attribute information. become. Therefore, it is possible to detect scenes automatically to some extent without relying on visual observation, and it is not necessary to decode all moving image information for detection, so that work time can be reduced.

Further, in the present invention, since a scene change is detected by using an encoding parameter, a scene change can be detected accurately.

For the detected candidate scene,
After decoding and displaying this candidate scene and the image information of several frames before and after, the user determines the type of the scene, so that not only one type of scene obtained as a detection result but a plurality of scenes is obtained. Can be detected. In addition, by recording the detection / determination result, that is, the detected scene type and frame position together with the moving image information as editing information, it is possible to reuse the edited scene as moving image information. .

[Brief description of the drawings]

FIG. 1 is a basic configuration diagram of a moving image scene detection device of the present invention.

FIG. 2 is a schematic configuration diagram showing an embodiment of a moving image scene detection device of the present invention.

FIG. 3 is a configuration diagram of code information in the embodiment of FIG. 2;

FIG. 4 is a principle diagram of motion compensation.

FIG. 5 is a block diagram showing a signal flow in the embodiment of FIG. 2;

6 is a diagram showing a flow of processing in a candidate scene detection unit in the embodiment of FIG.

FIG. 7 is a diagram illustrating a display example on an image display unit.

FIG. 8 is a diagram illustrating an example of using editing information.

FIG. 9 is a block diagram illustrating a moving image encoder for generating an encoded image on which scene detection is performed.

FIG. 10 is a block diagram showing another embodiment of the moving image scene detection device of the present invention.

FIG. 11 is a block diagram showing still another embodiment of the moving image scene detection device of the present invention.

FIG. 12 is a waveform chart showing an experimental result of scene change detection.

FIG. 13 is a schematic configuration diagram of a scene detection support device in a conventional example.

FIG. 14 is a diagram showing moving image information in a conventional example.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Instruction input / output part, 2 ... Information storage part, 3 ... Attribute information decoding part, 4 ... Candidate scene detection part, 5 ... Image information decoding part, 6 ...
Image information display section, 7 ... Detection recording start instruction signal, 8 ...
Attribute information, 9: composite attribute information, 10: attribute information read instruction signal, 11: designated frame image display instruction signal, 12
... image information, 13 ... decoded image information, 14 ... recording detection restart instruction signal, last frame instruction signal, 16 ... recording detection end instruction signal, 17 ... detection end instruction signal, frame position code, 19 ... inter-frame / inner code, 20 code amount code, 2
1 ... vector number code, 30 ... block extraction circuit, 31 ...
Subtracter, 32 discrete cosine transform circuit, 33 quantization circuit, 34 variable length coding unit, 35 inverse quantization circuit, 36
... Inverse discrete cosine transform circuit, 37 ... Adder, 38 ... Frame memory, 39 ... Significant / insignificant block control unit, 40 ...
Encoding mode control unit 41 41 first switch circuit 42 42
Second switch circuit, 51: variable length decoding unit, 52: decoding unit, 53: scene change detection filter, 54a, 54
b, 54c... statistic calculation circuit, 55a, 55b, 55c
... comparator, 56 ... AND gate, 101 ... information storage means, 102 ... attribute information decoding means, 103 ... candidate scene detection means, 104 ... image information decoding means, 105 ... image information display means

Claims (2)

[Claims] An information storage unit for storing code information including a moving image information main body, which is encoded moving image information, and attribute information indicating an attribute of the moving image information main body; and attribute information in the code information. Attribute information decoding means for decoding the attribute information, candidate scene detection means for detecting a candidate for a characteristic scene in a moving image using attribute information output from the attribute information decoding means, and image information in the code information. A moving image scene detection apparatus comprising: image information decoding means for decoding; and image information display means for displaying image information output from the image information decoding means. 2. The method according to claim 1, wherein the code information is coded using coding that does not perform code amount control for each frame, and the attribute information is a code amount for each frame. The moving image scene detection device according to claim 1.