JPH06339109A - Scene change detection method, its device and moving picture edit device - Google Patents

Abstract

(57) [Summary] [Purpose] A scene change detection method and device that can determine scene changes without being affected by movements inside the screen, and browsing that makes it easy to grasp the image contents, and movies that enable accurate tape positioning. To provide an image editing device. [Structure] A block position setting unit 11 that sets a fixed block and a moving block, a histogram calculation unit 12 that calculates a histogram of a fixed block of a previous frame and a histogram of a moving block of a subsequent frame, and a correlation value between those histograms is calculated. Correlation value calculating unit 14, a matching unit 15 that obtains the highest correlation value among the calculated correlation values for each fixed block, and based on a predetermined number of correlation values from the highest correlation value, the entire screen A full-screen correlation calculation unit 16 that calculates a correlation value and a scene change determination unit 17 that determines that there is a scene change based on the full-screen correlation value are provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a scene change detecting method and apparatus for detecting a frame in which a scene of a moving image greatly changes, and browsing using the frame of the scene change in editing a moving image such as a video tape. , A playback / recording / fast-forward / rewind for each scene, and automatic generation of a digest version of a moving image.

[0002]

2. Description of the Related Art With the spread of video cameras and the trend of multimedia in recent years, there are increasing opportunities for users to edit moving images such as video tapes and optical discs. For example, T
Editing includes creating a digest version of a V program, joining tapes that were shot by myself to only the necessary parts, changing the order, and dubbing to another tape.

Conventionally, as a method for automatically creating a digest version such as recording, a user judges the importance (priority) of each frame image, and a frame image having a priority higher than a threshold value is used as a digest version. A method of recording has been proposed (for example, JP-A-3-90968).

However, in order for the user to judge the priority of the image, the position of the tape is adjusted in order to know what kind of image is recorded on which part of the tape or to determine the range of the image to be recorded as a digest. It is necessary to do work such as doing, and conventional simple functions such as fast-forward, rewind, and playback are troublesome.

Therefore, there has been proposed a device that finds a break (hereinafter referred to as a scene change) in a moving image and simplifies the work of browsing and aligning the tape based on the break. For example, the moving image editing apparatus disclosed in Japanese Patent Laid-Open No. 2-184181 detects a scene change based on the illuminance, volume, color tone, etc. of the moving image, and the user inputs a button to fast-forward / play to the next scene or change the previous scene. It is rewinding up to.

Further, as a method for automatically detecting a scene change from a video signal, there is a method disclosed in, for example, Japanese Patent Laid-Open No. 3-214364. This is to compare histograms of luminance levels of all screens between adjacent frames and determine that there is a scene change when the sum of absolute values of histogram differences exceeds a predetermined value.

As another scene change detecting method,
There is also a method in which the screen is divided into a plurality of blocks and the histogram is compared not for the entire screen but for each block. (For example, Nagasaka, Tanaka, “Automatic indexing method and object searching method in color video images”, IPSJ Journal Vol.33, No.4 (1992)). The operation of the above method will be briefly described below.

FIG. 5 is a diagram showing the positional relationship of blocks for comparing histograms between adjacent frames, at time t =
Each frame of the image of τ and t = (τ + 1) is divided into 16 blocks. Hereinafter, these 16 blocks will be referred to as fixed blocks, and the Xth block from the left in the horizontal direction and the Yth block from the top in the vertical direction will be referred to as the (X, Y) fixed block (1≤X, Y≤. 4). The color histogram of the (X, Y) fixed block of the frame at time t is described as hist (t, X, Y, c). However,
The colors in the image are quantized into 64 color types, and hist (t,
X, Y, c) indicates the number of pixels having the c-th color in the block.

First, as shown in FIG. 5, the color histograms of the fixed blocks at the same positions of the two frames are compared. Here, as a histogram comparison method, χ2
Use a test. For the (X, Y) fixed block of the frame at time τ 2, the evaluation value kai (τ, X, Y) is obtained as follows.

[0010]

[Equation 1]

The evaluation values are obtained for 16 fixed blocks by (Equation 1), and then the sum of the eight evaluation values of the smaller 16 evaluation values is obtained, and the final evaluation value S is obtained.
For simplicity, the method of deriving the final evaluation value is written as (Equation 2).

[0012]

[Equation 2]

When the final evaluation value S thus obtained is larger than a predetermined threshold value, it is determined that the scene has changed between the frames of t = τ and t = τ + 1.

[0014]

In the above scene change determination, in order to detect just the frames with scene changes without excess or deficiency, image fluctuations other than scene changes, such as movement / deformation of the subject within the screen, are necessary. An evaluation value that can determine that the correlation between frames is high even if it occurs is necessary.

However, the conventional scene change detection method has the following problems.

On the other hand, in the method of comparing the histograms of the entire screen between adjacent frames, it is determined that the correlation is high unless the histogram of the entire screen is changed. There is a wide range not to be done. However, if there is a change in the area of the subject or a new background emerges from the place where the object existed, even if it is a part of the screen, it will affect the histogram of the entire screen to some extent. As a result, there are cases where it is erroneously determined that there is a scene change even if there is a local image change.

On the other hand, in the method of comparing the histograms for each of the blocks obtained by dividing the screen, only blocks having a high correlation are selected and the final evaluation value is calculated. It is excluded and is not affected by it. However, since the allowable range of object movement is limited to the inside of the block, for example, when an object occupying a large area in the screen moves across different blocks,
Even if the moving distance is small, the number of blocks having low correlation increases, so that it may be erroneously determined that there is a scene change.

As described above, in the conventional scene change detection method, since it is erroneously detected that there is a scene change even in the case of an image change, a moving image editing apparatus using the scene change as a unit does not require browsing or tape alignment. There is a problem that it cannot be done accurately.

The present invention takes such problems of the conventional scene change detection method into consideration, and a scene change detection method and its apparatus capable of determining a scene change without being influenced by the movement inside the screen, and grasping the image contents. An object of the present invention is to provide a moving image editing apparatus that enables easy browsing and accurate tape position adjustment.

[0020]

According to the present invention of claim 1, the first frame image is divided into a plurality of first fixed blocks,
Also, the second frame image is divided into second fixed blocks in the same manner as the first fixed block is divided, and each second fixed block is a moving block having the same size as the second fixed block. Are set in a predetermined area near the second fixed block so as to overlap or not overlap with the second fixed block, and predetermined image information regarding the image in the first fixed block is calculated. The image information of each of the plurality of moving blocks regarding the second fixed block corresponding to the block is calculated, and the correlation value between the image information of the first fixed block and the image information of the moving block is calculated, and the calculation is performed. The maximum of the correlation values
This is a scene change detection method that determines the block correlation value of the first fixed block, obtains the block correlation value for all the first fixed blocks, and determines whether or not there is a scene change in the frame image based on the block correlation value.

According to a fifth aspect of the present invention, the first frame image is divided into a plurality of first fixed blocks, and the second frame image is divided into the first fixed blocks in the same manner as the first fixed block. It is divided into two fixed blocks, and for each of the second fixed blocks, a plurality of moving blocks of the same size as the second fixed block are provided in a predetermined area near the second fixed block.
A block setting means for setting the fixed block so as to overlap or not overlap with the fixed block, predetermined image information about an image in the first fixed block, and a plurality of second fixed blocks corresponding to the first fixed block. Histogram calculating means for calculating each image information in the moving block, and a correlation value between the image information of the first fixed block and the image information of the moving block are respectively calculated, and the maximum correlation value among the calculated correlation values is calculated. A block correlation value calculating means for obtaining a block correlation value for all the first fixed blocks, the value being a block correlation value of the first fixed block,
The scene change detecting device includes a scene change determining unit that determines whether or not there is a scene change in a frame image based on the block correlation value calculated by the block correlation value calculating unit.

According to a sixth aspect of the present invention, there is provided a moving image storage means for storing a frame image of a moving image, a scene change detection device for detecting a scene change of the stored image frame, and the detected scene change device. In a moving image editing apparatus including a scene change storage unit that stores frame position information of a moving image in which a scene change has occurred, and a moving image display unit that displays a frame image according to the stored frame position information. is there.

[0023]

According to the present invention, the predetermined image information regarding the image in the first fixed block is calculated, and the respective image information within the plurality of moving blocks regarding the second fixed block corresponding to the first fixed block is calculated. Then, the correlation value between the image information of the first block and the image information of the moving block is calculated, and the maximum value among the calculated correlation values is calculated as
The block correlation value of the first fixed block is set, the block correlation value is calculated for all the first fixed blocks, and it is determined that there is a scene change in the frame image based on the block correlation value. The influence of image fluctuation can be suppressed.

According to the present invention, the moving image storing means stores the frame image of the moving image, the scene change detecting device detects the scene change of the stored image frame, and the scene change storing means detects the detected scene change. The frame position information of the generated moving image is stored, and the moving image display means displays the frame image according to the stored frame position information.

[0025]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below with reference to the drawings showing its embodiments.

FIG. 1 is a diagram showing the functional blocks of a scene change detection apparatus according to an embodiment of the present invention, and FIG.
FIG. 7 is a diagram showing a positional relationship of blocks for comparing histograms between adjacent frames in the present embodiment (see FIG.
3), and FIG. 3 is a diagram in which the center position of the block for which the histogram calculation is performed in the present embodiment is shown by a black dot.

First, the scene change detection method of this embodiment will be described with reference to FIG. This method differs from the conventional method in that each fixed block of a frame of t = τ and t = (τ +
Correlation calculation is performed between a plurality of blocks around each fixed block in the frame 1). However, the periphery here means an area expanded by one block in the vertical and horizontal directions from each fixed block. For example, FIG. 2 shows an example of a block in which a (2,3) fixed block of a frame of t = τ performs correlation calculation in a frame of t = (τ + 1) by a dotted-line rectangle. Here, a large rectangle with a thick line is a peripheral region of the (2,3) fixed block and indicates a range in which the block can exist.

At this time, since it is not necessary to calculate the correlation between the fixed block and all the blocks that may exist within the above range, the blocks used for the correlation calculation are thinned out at a predetermined interval. Here, sample points sampled at intervals of ⅕V and ⅕H (H and V are the horizontal and vertical lengths of one block, respectively) in the horizontal and vertical directions within the existing range are located at the center position. Select a block. For example, the block corresponding to the (2,3) fixed block in FIG.
The number is (= 11 × 11). After that, the correlation calculation with the (X, Y) fixed block of the frame of t = τ is performed, and t = (τ +
The block of the frame 1) is called a moving block related to the (X, Y) fixed block.

As described above, each fixed block of the frame of t = τ is subjected to correlation calculation with respect to all moving blocks related to each fixed block, and the correlation value with the moving block having the highest correlation among them is calculated as the fixed block. The evaluation value of Subsequent processing is the same as in the conventional method.
A final evaluation value is obtained based on the evaluation values of the fixed blocks, and the final evaluation value is used for the scene change determination.

Next, an example of the structure of a scene change detecting apparatus for realizing the scene change detecting method of the present invention will be described with reference to FIG.

In FIG. 1, 10 is a video signal input terminal to which an input image signal is input, 11 is a block position setting section for outputting the position of a block for histogram calculation, and the coordinates indicated by black dots in FIG. The data is sequentially output at predetermined time intervals. However, the black dots in Fig. 3 are 1 in the horizontal and vertical directions.
It is sampled at intervals of / 5H and 1 / 5V, and corresponds to the center coordinates of the block. Hereinafter, the block at the position set by the block position setting unit is called a mask block (that is, the mask block is a block set corresponding to the positions of the fixed block and the moving block).

Reference numeral 12 is a histogram calculation unit for calculating a color histogram of the mask block designated by the block position setting unit 11 from the input video signal.

Reference numeral 13 denotes a histogram accumulating section for storing the histogram in the fixed block calculated by the histogram calculating section 12. For each fixed block, the memory (A) for the immediately preceding frame and the temporary frame of the frame currently being processed are stored. It has two memories, a storage memory (B).

Reference numeral 14 is a histogram in the fixed block of the immediately preceding frame stored in the histogram storage unit 13,
It is a correlation value calculation unit that performs correlation calculation with the histogram of the moving block calculated by the histogram calculation unit 12. Corresponding to each fixed block is composed of 16 units, and only the unit whose mask block is within the block existence range around each fixed block performs the correlation calculation. Here, the correlation value is calculated using the sum of absolute values of the histogram differences.

Reference numeral 15 denotes a matching unit that outputs the correlation value (block correlation value) between the fixed block and the mask block having the highest correlation among the correlation values calculated by the correlation value calculation unit 14. Corresponding to 16
It consists of individual units.

Reference numeral 16 designates 8 correlation values having the highest correlation among 16 correlation values (block correlation values) input from the matching section 15 and outputs the sum as a full screen correlation value. It is a screen correlation calculation unit.

Reference numeral 17 is a scene change determination unit that compares the output of the full-screen correlation calculation unit 16 with a predetermined threshold value and determines whether there is a scene change, and 18 is a scene change determination unit 17 from the scene change determination unit 17. This is an output terminal for outputting the processing result.

Here, the block position setting unit 11 described above is a block setting unit, the histogram calculation unit 12 and the histogram accumulation unit 13 constitute a histogram calculation unit, and the correlation value calculation unit 14 and the matching unit 15 are block correlation values. The calculation unit is configured, and the full-screen correlation calculation unit 16 and the scene change determination unit 17 configure the scene change determination unit.

Next, the operation of the above embodiment will be described.

Now, assuming that the video signal relating to the frame of t = τ is inputted from the input terminal 10, the frame memory (A) immediately before the histogram accumulating section 13 is
It is assumed that the histogram in the fixed block for the frame of t = (τ−1) is recorded.

First, the block position setting unit 11 selects the upper left black point in FIG. 3 and outputs its coordinates (x0, y0).
The histogram calculation unit 12 sets a mask block whose center coordinates are (x0, y0) using the input video signal, and calculates a color histogram inside the block.

Next, in the histogram accumulation section 13, the mask block position (x0,
y0) and check if the position is equal to any of the 16 fixed blocks. When there is a fixed block with the same position, the histogram output from the histogram calculation unit 12 is recorded in order to update the contents of the temporary storage memory (B) corresponding to the fixed block. Now, since the central coordinates (x0, y0) of the mask block match the central coordinates of the (1,1) fixed block, the temporary storage memory (B) of the histogram accumulation unit 13 corresponding to the (1,1) fixed block. The content of is updated.

Next, the correlation value calculation unit 14 first receives the mask block position (x0, y0) from the block position setting unit 11.
, And select a fixed block whose mask block is included in the surrounding block existence range. When the mask block position is (x0, y0), the fixed block including the mask block in the block existence range is (1, 1),
There are four (1,2), (2,1), and (2,2). The unit corresponding to the selected fixed block is the histogram of the mask block from the histogram calculation unit 12 and the frame memory (A) immediately before the histogram storage unit 13.
A fixed block histogram is input from and the correlation value between the two histograms is obtained.

Then, the matching unit 15 stores the correlation value input from the correlation value calculation unit 14 for each fixed block.

When the above processing is completed, the block position setting section 11 selects another black point and outputs its coordinates (x1, y0) as shown in FIG. The histogram calculation unit 12, the histogram accumulation unit 13, and the correlation value calculation unit 14 are (x0, y
Perform the same processing as in (0). Then, the matching unit 15 compares the input correlation value with the already stored correlation value, and stores only the larger correlation value. The unit to which the correlation value is input for the first time stores the correlation value.

As described above, with respect to the coordinates of the mask block output by the block position setting unit 11, the histogram calculation unit 12, the histogram storage unit 13, and the correlation value calculation unit 1
4. The processing of the matching unit 15 is performed one after another. As a result, when the processing for all the black dots in FIG. 3 is completed,
In the 16 units of the matching unit 15, the correlation value between the fixed block and the mask block having the highest correlation is stored for each fixed block.

When the processing for all the black dots is completed, the block position setting section 11 sends a signal to the histogram accumulating section 13 and the matching section 15 to notify the end of the processing. When the processing end signal is input, the histogram storage unit 13 updates the contents of all the previous frame memories (A) with the corresponding values of the temporary storage memory (B).

On the other hand, the matching section 15 outputs the correlation value for each fixed block to the full screen correlation calculating section 16, and the full screen correlation calculating section 16 selects eight correlation values from the one having the highest correlation. , The sum is output as the full screen correlation value.

Then, in the scene change judging section 17,
A predetermined threshold value is compared with the full-screen correlation value output from the full-screen correlation calculation unit 16. When the full-screen correlation value is smaller than the threshold value, it is determined that there is a scene change, and the determination result is output from the output terminal 18. Output.

The above is the processing for one frame. The same processing is performed for the subsequent frames, and the determination result as to whether or not there is a scene change is sequentially output.

As described above, when investigating the correlation of the histograms in a block, the correlation of the histogram is calculated not with the blocks at the same position but with the block having the highest correlation in the vicinity, so that the vicinity of the block boundary is determined. Therefore, even if the subject moves, it can be determined that the correlation is high. Therefore, it is possible to extend the range in which the subject can move without being determined to have a scene change, while suppressing the influence of local image variation, and to reduce the local change of the scene change detection method using the conventional in-block histogram. False detection of scene changes, while maintaining the advantage of being less affected by image variations
The detection omission can be reduced. Further, this makes it possible to accurately detect a scene change when performing editing work, and improve the efficiency of editing work such as browsing and tape position alignment.

In the above embodiment, the sum of absolute values of histogram differences is used as the histogram correlation calculation method. However, the present invention is not limited to this. For example, a χ2 test or the like may be used as in the conventional method. Good.

Further, in the above embodiment, the correlation calculation is performed using the histogram of the color level as the image information, but the present invention is not limited to this, and a histogram of the brightness level or the like may be used.

Further, in the above embodiment, the range in which the blocks in the peripheral area of the fixed block, which is the predetermined area near the fixed block, can exist is the range constituted by the fixed blocks adjacent to the fixed block (see FIG. 2). However, the range of the area is not limited to this.

In the above embodiment, the moving blocks are set at the intervals of 1 / 5V and 1 / 5H in the horizontal and vertical directions, respectively, but the interval is not limited to this.

Further, in the above-described embodiment, half (8) of all (16) block correlation values are used to obtain the full screen correlation value, but the present invention is not limited to this, for example, all or 6. Any number such as an individual may be used.

Although the screen is divided into 16 blocks in the above embodiment, the number of divisions is not limited to this.

In the above embodiment, the fixed block histogram is calculated from the previous frame, and the moving block histogram is calculated from the subsequent frame. Conversely, the fixed block histogram is calculated from the subsequent frame. Alternatively, the histogram of the moving block may be calculated from the previous frame.

Further, in the above embodiment, the block correlation value of the first fixed block is the maximum value among the obtained correlation values between the image information of the first fixed block and the image information of the moving block. However, the present invention is not limited to this,
It may be a value obtained by another method such as an average value of the maximum value and the second largest value.

Further, in the above-mentioned embodiment, although the configuration is such that adjacent frames are compared with each other, the present invention is not limited to this, and it is also possible to compare several frames apart from each other.

Further, in the above-mentioned embodiment, each processing unit is constituted by dedicated hardware, but instead of this, the same function may be realized by software using a computer.

Next, an embodiment of a moving image editing apparatus using the scene change detecting method (or apparatus) of the present invention shown in FIG. 4 will be described. The moving image editing apparatus of this embodiment is
By detecting a scene change from a video signal of a moving image and displaying an image representing a lump of moving images between the scene changes, browsing work is facilitated.

In FIG. 4, the moving image storage unit 1 is a moving image storage means composed of a video tape, an optical disc, or the like, and stores video signals and synchronizing signals of moving images. The operation input unit 2 is an input device for a user to input a command for work such as browsing start and scene change detection start. The image output unit 3 is a moving image display unit that displays an image on an output device such as a display. The scene change detection unit 5 is a scene change detection device that detects a scene change from a video signal of a moving image in the moving image storage unit 1 using the scene change detection method of the present invention. The scene change storage unit 6 is a scene change storage unit that stores the beginning frame number of the scene based on the processing result from the scene change detection unit 5. The control unit 4 controls the exchange of signals among the moving image storage unit 1, the operation input unit 2, the image output unit 3, the scene change detection unit 5, and the scene change storage unit 6 described above.

Next, the operation of the moving image editing apparatus of the above embodiment will be described.

When the operation input section 2 detects a button input signal for starting the scene change detection, all the scene changes in the moving image storage section 1 are detected and stored in the scene change storage section 6. Begins.

First, the control unit 4 transmits the video signal of the moving image storage unit 1 to the scene change detection unit 5. The scene change detection unit 5 detects a scene change using the scene change detection method of the present invention described above, and the frame number at the beginning of the scene is recorded in the scene change storage unit 6 based on the detection result. The above processing is performed for all the frames stored in the moving image storage unit 1 or the frames in the specified range, and the processing is performed when all the scene changes therein are stored in the scene change storage unit 6. finish.

In this state, when the operation input section 2 detects a button input signal for starting browsing, processing for displaying the image at the beginning of the scene on the display is performed. First, the control unit 4 refers to the frame number recorded in the scene change storage unit 6, controls the moving image storage unit 1 to search for the corresponding frame, and transmits the video signal of the frame to the image output unit 3. . The image output unit 3 is set so that a plurality of images can be simultaneously displayed in the form of an icon, and the transmitted image is displayed on the display.

As described above, by displaying the image at the beginning of the scene based on the change point obtained by the scene change detection method of the present invention, the image can be viewed without excess or deficiency for each scene, and the moving image can be viewed. It becomes possible to grasp the contents of efficiently and easily. That is, it is possible to perform efficient editing without problems such as displaying the image of the same scene many times during browsing, skipping the changing point of the scene by fast-forwarding to the beginning of the next scene, and the like.

In the above embodiment, it was explained that only the image at the beginning of the scene is displayed, but the present invention is not limited to this.
Select a frame that has passed a predetermined time from the beginning of the scene,
It may be displayed by selecting the middle frame during the scene.

In the above embodiment, the method of simultaneously displaying all the scene changes in the form of icons has been described, but the present invention is not limited to this, and a plurality of frames are selected for each scene and sequentially displayed like a digest. You may do it.

Further, in the above-described embodiment, the image representing the scene is displayed for browsing. However, the present invention is not limited to this, and the configuration similar to that of FIG. It may be configured to perform tape position adjustment work such as reproduction and fast-forward to the scene.

[0072]

As is apparent from the above description, the present invention has an advantage that a scene change can be determined without being influenced by the movement inside the screen.

Further, since the present invention is a moving image editing apparatus equipped with a scene change detecting device for detecting a scene change of an image frame, there is an advantage that it is possible to browse the image contents easily and perform accurate tape position alignment. is there.

[Brief description of drawings]

FIG. 1 is a functional block diagram of a scene change detection apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram showing a positional relationship of blocks for comparing histograms between adjacent frames in the embodiment.

FIG. 3 is a diagram showing the center position of a block for which histogram calculation is performed in the same embodiment as a black dot.

FIG. 4 is a block diagram showing a configuration of a moving image editing apparatus according to an embodiment of the present invention.

FIG. 5 is a diagram showing a positional relationship of blocks for comparing histograms between adjacent frames in a conventional scene change detection method.

[Explanation of symbols]

 1 Moving image storage unit 2 Operation input unit 3 Image output unit 4 Control unit 5 Scene change detection unit 6 Scene change storage unit 11 Block position setting unit 12 Histogram calculation unit 13 Histogram storage unit 14 Correlation value calculation unit 15 Matching unit 16 Full screen Correlation calculation unit 17 Scene change determination unit

Claims (6)

[Claims] 1. A first frame image is divided into a plurality of first fixed blocks, and a second frame image is divided into second fixed blocks in the same manner as the first fixed block is divided. Then, for each second fixed block, the second
A plurality of moving blocks having the same size as the fixed block are set in a predetermined area near the second fixed block so as to overlap or not overlap with the second fixed block.
Predetermined image information about the image in the fixed block is calculated, and the image information of each of the plurality of moving blocks related to the second fixed block corresponding to the first fixed block is calculated, and the first image information is calculated. The correlation value between the image information of the fixed block and the image information of the moving block is calculated, and the first correlation value is calculated from the calculated correlation value.
Scene change detection, characterized in that a block correlation value of a fixed block is obtained, the block correlation value is obtained for all the first fixed blocks, and whether or not there is a scene change of the frame image is determined based on the block correlation value. Method. 2. The block correlation value of the first fixed block is
The scene change detection method according to claim 1, wherein the scene change detection method is a maximum value among the obtained correlation values between the image information of the first fixed block and the image information of the moving block. 3. After obtaining a block correlation value for all of the first fixed blocks, a predetermined number is extracted from the higher one of the block correlation values, and based on the extracted block correlation value, a whole screen is extracted. 3. The full-screen correlation value representing the correlation value is obtained, and when the full-screen correlation value is lower than a predetermined threshold value, it is determined that there is a scene change of the frame image. Scene change detection method. 4. The scene change detection method according to claim 1 or 2, wherein the predetermined image information is a color signal or a luminance signal. 5. The first frame image is divided into a plurality of first fixed blocks, and the second frame image is divided into second fixed blocks in the same manner as the first fixed block is divided. Then, for each second fixed block, the second
Block setting means for setting a plurality of moving blocks having the same size as the fixed block in a predetermined region near the second fixed block in a state of overlapping or not overlapping with the second fixed block, and in the first fixed block. Histogram calculation means for calculating predetermined image information about an image, and calculating the image information of each of the plurality of moving blocks for the second fixed block corresponding to the first fixed block, The correlation value between the image information of the fixed block and the image information of the moving block is calculated, and the maximum value among the calculated correlation values is set as the block correlation value of the first fixed block. A block correlation value calculation means for obtaining a value for all the first fixed blocks, and a block correlation value calculation means for obtaining the block correlation value calculation means. Based on the click correlation value, the scene change detecting device characterized by comprising a scene change judging means for judging whether said frame image is a scene change. 6. A moving image storage means for storing a frame image of a moving image, the scene change detection device according to claim 5 for detecting a scene change of the stored image frame, and the detected scene change occurs. And a scene change storage unit for storing frame position information of the moving image and a moving image display unit for displaying the frame image according to the stored frame position information. .