JP2003047004A - Image feature detecting method, image feature detecting equipment and data recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately distinguish a flashing image from scene change points in a video signal with little data and detect features. SOLUTION: An inputted image plane is divided into four regions by an image plane dividing means 101, and picture element data in the regions are stored by a region storing means 102. A parameter calculating means 103 obtains a sum of sign added differential values of picture element values of the region stored by the storing means 102 and of a region which is obtained by dividing the next field image plane. A pictorial image plane detecting means 104 compares obtained parameters with a threshold group, and detects features of a pictorial image and an image plane.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video feature detection method, a video feature detection device, and a data recording medium,
In particular, the present invention relates to a flashing image in a video signal whose brightness changes drastically and a function capable of accurately detecting a scene change point at which temporal continuity is lost.

[0002]

2. Description of the Related Art In recent years, digital signal processing technology has been developed,
Digital compression of moving images is becoming commonplace.
In particular, to realize high compression coding, MPEG (Mov
ingPicture Experts Group)
The interframe predictive coding process used in such as is essential. In the inter-frame predictive coding, high-compression coding of a moving image is realized by expressing information about the motion of an object using a motion vector.

However, a motion vector obtained without special processing between video images whose brightness changes drastically, for example, a video image that crosses a scene change point where the scenes of the video are switched, is a pixel of the original screen. A suitable one was detected only in the luminance level of. Therefore, a motion vector that is completely different from the motion of the object on the screen is often detected, or even if the target object disappears when the scene switches and the motion vector is often forcibly detected,
There is a problem that an image compressed using this motion vector becomes an image that is noticeably deteriorated as compared with the original image when viewed by human eyes.

Therefore, it is possible to detect a motion vector with less deterioration by changing the method of detecting the motion vector according to the characteristics of the image or screen. For this purpose, a method for detecting the characteristics of video and screen is required, and several methods have been proposed. For example, Patent Publication No. 2
No. 611607 and Japanese Patent Publication No. 2565057 are listed. In Japanese Patent No. 2611607, the input image is divided into a plurality of small screens, and a parameter calculated between temporally continuous small screens is compared with a threshold to detect a scene change. Has been realized.

Further, in Japanese Patent Publication No. 2565057,
A scene change can be detected by setting a flag that indicates the possibility of a scene change from the data obtained from the image differences between the screens that are continuous in time and seeing the appearance pattern of multiple flags that are continuous in time. is doing.

However, in the case of an image in which the images whose brightness changes drastically are continuous, for example, in the case of a flashing image in which the flash of the camera is hit or the image of which the illumination flashes, this is the scene. There is a problem that it is erroneously determined as a change. More specifically, in the above-mentioned Japanese Patent No. 2611607, the parameter to be compared according to the threshold value is only the data between two temporally consecutive screens, and the temporally preceding and subsequent images are considered. Therefore, in an image in which the brightness continuously changes drastically like flashing, there is a possibility that it may be erroneously detected as a continuous scene change. Further, in the above-mentioned Japanese Patent Publication No. 2565057, since the input screen is used as it is, it is impossible to detect the flashing that may appear as a local luminance change of the screen.

[0007]

The conventional image feature detecting method and image feature detecting device are configured as described above.
No consideration is given to a flashing image in which the brightness continuously and drastically changes, and therefore, a scene change cannot be detected accurately, and as described above, high compression encoding of a moving image using a motion vector is performed. However, there is a problem in that the problem that the deterioration of the image becomes noticeable cannot be solved.

The present invention has been made in order to solve the above-mentioned problems, and it detects the characteristics of the image and the screen with high accuracy, does not have a temporal continuity, and has a scene change and a temporal continuity. It is an object of the present invention to provide a video feature detection method, a video feature detection device, and a data recording medium, which are capable of detecting flashing each having a property.

[0009]

According to a first aspect of the present invention, there is provided a video feature detecting method for detecting a video feature or a temporal feature of a screen on a screen of images continuously input in time. A detection method, a screen dividing step of dividing a screen of an input video into a plurality of areas, an area storing step of storing image data of the divided screen area, and an area stored in the area storing step. Of the image data of the stored area and the image data obtained by dividing the image of the area temporally continuous with the stored image data of the area into parameters for calculating pixel data between these two pieces of image data. From the calculation step and the parameters obtained in the parameter calculation step, the change in the pixel data corresponds to the time defined by the first threshold value group. A video screen feature detection step of determining whether or not generated, is intended to include.

According to a second aspect of the present invention, in the image feature detecting method according to the first aspect, the change in the pixel data corresponds to the time defined by the first threshold value group. The step of determining whether or not the error has occurred includes an area-specific determination step of making a determination in each of the divided areas.

The video feature detecting method according to claim 3 of the present invention is the video feature detecting method according to claim 1, wherein the video screen feature detecting step is performed at a time defined by the first threshold value group. It includes a step of determining whether or not the value calculated from the parameter obtained in the corresponding period corresponds to the value defined by the second threshold value group.

According to a fourth aspect of the present invention, there is provided the image feature detecting method according to the third aspect, wherein the image feature detecting method is obtained in a period corresponding to the time defined by the first threshold value group. The image feature detecting method, wherein the value calculated from the parameter is the difference between the maximum value and the minimum value of the parameter obtained during the period corresponding to the time defined by the first threshold value group.

According to a fifth aspect of the present invention, there is provided the image feature detecting method according to the second aspect, wherein the parameter of each area which is the determination target in the area-specific determination step is the third parameter. The method includes a threshold value number detecting step of detecting whether or not the number corresponding to the value set in the threshold value group is equal to the number set in the fourth threshold value group.

The image feature detecting method according to claim 6 of the present invention is the image feature detecting method according to claim 5, wherein a predetermined number of screens temporally precedes and follows the screen of the pixel data which is the determination target. For the pixel data of, the value of the parameter of each area obtained by dividing the screen is compared with the value corresponding to the value defined by the third threshold value group, and the comparison result is compared with the fourth threshold value. It includes the step of comparing with the number defined by the value group.

The image feature detecting method according to claim 7 of the present invention is the image feature detecting method according to claim 1, wherein the calculated parameter is the time-dependent value for each pixel in the divided screen area. It is the sum of the signed difference values of the pixel values that change with time.

The video feature detection method according to claim 8 of the present invention is the video feature detection method according to claim 5, wherein the sum of the signed difference values of the pixel values is within the divided screen area. The pixel difference calculation step of calculating a signed difference value over time for each pixel, and the area sum calculation step of calculating the sum of the values calculated in the pixel difference calculation step within the area. is there.

The image feature detecting method according to claim 9 of the present invention is the image feature detecting method according to claim 7, wherein the sum of the signed difference values between the pixel values is within the divided screen area. The calculation is performed by a pixel sum calculation step of calculating the sum of pixel values and an area difference calculation step of calculating a signed difference value over time from the values calculated in the pixel sum calculation step.

The video feature detecting apparatus according to claim 10 of the present invention is a video feature detecting apparatus for detecting a video or a temporal feature of a screen in a screen of images continuously input in time. Screen dividing means for dividing the screen of the input video into a plurality of areas, storage means for storing the image data of the divided screen areas, image data of the areas stored by the storage means, and From the image data of the stored region image data and the image data obtained by dividing the image of the region continuous in time by the screen dividing means,
From the parameter calculation means for calculating the parameter relating to the pixel data between these two image data and the parameter obtained by the parameter calculation means, the change of the pixel data corresponds to the time defined by the first threshold value group. And a video screen feature detecting means for determining whether or not it has occurred.

A data storage medium according to an eleventh aspect of the present invention is a video feature detection program for causing a computer to perform a process of detecting a temporal feature of a video or a screen which is continuously input in time. And a first program for causing a computer to perform a screen division process for dividing the screen of the input video into a plurality of areas. A second program for performing a process of storing image data of a screen area by a computer, image data of an area stored in the second program, and an area temporally continuous with the image data of the stored area From the image data obtained by dividing the image of the above image into screens by the above-mentioned first program, The process of calculating the parameters relating to chromatography data, and a third program for a computer,
A video screen feature detection process for determining whether or not the change in the pixel data has occurred for a time period defined by the first threshold value group from the parameters obtained in the third program is performed by a computer. And a fourth program executed by.

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. (Embodiment 1) FIG. 1 is a block diagram showing the configuration of a video feature detection apparatus according to Embodiment 1 of the present invention. Figure 2
FIG. 2 is a diagram in which a flowchart showing a flow of processing performed by the image feature detection device shown in the block diagram of FIG. 1 is described. FIG. 3 is a schematic diagram showing an example of a divided screen obtained by the screen dividing means constituting the image feature detection apparatus, and shows an example when the screen is divided into four, and each divided area is divided into blocks. 1, block 2, block 3,
It is block 4. FIG. 4 is a schematic diagram for explaining an example of obtaining a difference between adjacent field screens. Further, FIG. 5 is a diagram for explaining the temporal change of the sum of the signed difference values of the block 1 which is one of the four divided areas.

In FIG. 1, 101 is an image dividing means for dividing the screen of the input image into a plurality of areas, 102 is an area storing means for storing pixel data in the divided areas, and 103 is the area storing means. Parameter calculation means for calculating a predetermined parameter from the pixel data of the area stored in 102 and the pixel data of the area obtained from the next field by the image dividing means 101, and 104 is the parameter obtained by the parameter calculation means 103. It is a video screen feature detecting means for detecting the feature of the video based on This example shows an example in which the screen of the input video is a field.

In more detail, the screen dividing means 1 shown in FIG.
The screen input with 01 is divided into several regions (here, four) as shown in FIG. The pixel data in these areas are stored in the area storage means 102. The area stored in the area storage means 102 and the screen division means 101
In the parameter calculation means 103, the sum of the sum of the signed difference values of the pixel value (here, the brightness is handled) with the pixel data at the corresponding position in the divided area of the next field screen in the same block. obtain. Specifically, as shown in FIG. 4, the signed difference value of pixel data at corresponding positions between adjacent field screens is calculated, and the signed difference value obtained between each frame is summed in the same block. Is taking. This process includes steps S201 to S201 in FIG.
This corresponds to the preprocessing portion of the flowchart in which the input is in field units and the output is in block units in 202. It is assumed that by this processing, in block 1, the sum of the signed difference values as shown in FIG. 5B described later is obtained.

At this time, the video screen feature detecting means 104 performs detection according to the following procedure. The flushing is detected for each block (here, four) into which the screen is divided. This process is a part of the flushing detection process in the flowchart of FIG. 2 (step S203, step S204).
Equivalent to. FIG. 5A and FIG. 5B are divided into 1
It shows changes in values over time in one region (block 1). That is, in FIG. 5A, the sum of the pixel values in the divided areas is displayed by numbering with the name DATA in the order of temporal flow. Figure 5
FIG. 5B shows a signed difference between temporally consecutive blocks with respect to the sum of pixel values in one block in FIG. 5A, and is numbered by the name DIFF in the order of temporal flow. I am wearing. For example, DIFF (n-5) is DATA (n-5)
Is obtained by subtracting DATA (n-6) from

A method of determining flushing and scene change will be described using the signed difference value of the pixel data of one block obtained as described above. Figure 5 (b)
As shown in, the value outside the section surrounded by the threshold value ± a is used as an effective value. This threshold value is set by experiment. If this threshold value is not set, the sensitivity of the device becomes too high, and the flashing or It will be determined whether or not there is a scene change. Although the threshold value is set to ± a for simplification of description here, the absolute values of the upper limit threshold value and the lower limit threshold value do not necessarily have to be the same value.

In FIG. 5 (b), the section of time from the DIFF (n-4) th to the DIFF (n-3) th which takes a + (positive) value is within the threshold value c, and In (a), DATA, which is the sum of the pixel values in the block in that section
(N-5) th, DATA (n-4) th, DATA (n-3) th
(Maximum value-minimum value) (here, DATA (n-
3) -DATA (n-5)) is greater than or equal to the threshold value d,
The block is detected as a scene that tends to be bright in flash. If the threshold value c is 3 and the sum (maximum value-minimum value) of the pixel values in the block in this section is greater than or equal to the threshold value d, in this example, the threshold value ± a is DIFF (n-4) th to DIFF (n-
Up to 3) th, this section is within the threshold value c, and DA
TA (n-5) th, DATA (n-4) th, DATA (n-
Since the (maximum value-minimum value) at the 3) th is equal to or greater than the threshold value d, this block is detected as a scene in which flash tends to be bright in this section.
Similarly, in the case of a negative (-) value in FIG. 5A, the scene is detected as a scene in which the flash tends to be dark.

In step S205, it is determined whether all the blocks have been subjected to the above processing, and steps S203 to S are performed until all the blocks are processed.
205 is repeated.

On the other hand, the scene change is performed in step S20.
In 6, the data for the entire screen is used to detect each field. That is, FIG. 6 also shows changes in values over time for each of the divided areas. As shown in FIG. 6, the DIFF data number when the block exceeds the section surrounded by the threshold ± b is viewed. Regarding the threshold value ± b at this time, the absolute values of the upper limit and the lower limit do not necessarily have to be the same value. At a certain DIFF data number, the number of blocks that exceed the section surrounded by the threshold ± b is equal to or greater than the threshold number of blocks, and the threshold two ± 2 before and after the threshold ± b. If the number of blocks in the enclosed section is equal to or greater than the threshold number of blocks, it is detected that the scene has changed in the field. By observing the data before and after the DIFF data that exceeds the threshold number in this way, the change in the pixel value of the pixel data is caused locally or on the entire screen, and By detecting whether or not they occur consecutively in time, it is possible to prevent erroneous detection of a scene change as flushing. In this example, the number of threshold blocks is three or more, and in FIG. 6, there are three DIFF (n−2) th blocks exceeding the section surrounded by the threshold ± b. Also,
All of the DIFF data within the section surrounded by the threshold ± b in the two DIFF data numbers before and after the DIFF data number. Therefore, it is detected that a scene change is performed between the fields in which the DIFF (n-2) th data is obtained.

Further, in the flowchart in FIG. 2, when flushing is detected in step S204 and a scene change is detected in the subsequent step S206, it means that flushing and scene change occur at the same time. Feature detecting means 104
The following determination is performed using the determination means provided inside. That is, when the video screen feature determination process shown in the flowchart of FIG. 2 is set to give priority to the scene change, if flushing and scene change are detected at the same time, even if there is an area determined to be flushing, The final decision is made that the scene has changed in the frame.

As described above, in the present embodiment, the sum of the signed difference values of the temporal pixel data of each pixel in the divided area of the screen of the input video is used as one parameter. Since the features of each image and the features of each region within the screen of the image are detected using the pattern of temporal change for each region, a scene that is an accurate feature of the image or screen with less data can be detected. Both change and flushing can be detected.

In the present embodiment, the screen division is simply divided into four, but it is also possible to increase the number of divisions, change the shape of the divided area, or use another division method. . Further, although the case where the input video is in field units has been described as an example in the present embodiment, the same effect can be achieved in frame units.

Further, as shown in FIG. 5A, the threshold value of the temporal change of the sum of the pixel values of the pixel data of one block, and between the blocks shown in FIG. 5B. As a threshold for the temporal width of the signed difference value, a single threshold is provided to classify the calculated data, but a plurality of thresholds are set here. A threshold value may be used to increase the number of data classifications.

Further, in the present embodiment, the threshold value is set for the temporal change amount of the sum of pixel values in one block shown in FIG. When it is not necessary to know, the value of this threshold is 0 (zero), that is, the threshold for the change amount does not have to be set, and the value of + (positive) shown in FIG. It is also possible to collectively detect the section that takes a value and the section that takes a value of- (negative) as flashing.

Further, a section which takes a value of + (positive), or
Depending on the setting of the threshold value group used in the section that takes a negative value, it can be detected as a fade-in in which the screen slowly becomes bright or a fade-out in which the screen becomes slowly dark.

Further, when detecting a scene change, two DIFF data each before and after the DIFF data number exceeding the threshold value were used for reference of determination, but only the previous or subsequent data may be used. , The number of data used is continuous 2
It is not limited to one data.

Further, when the flushing and the scene change are detected at the same time, the scene change is prioritized and determined here, but the flushing may be prioritized and determined by the user. In response to the,
When you can obtain an image with no noticeable deterioration,
Both can be set to be adopted or not adopted.

In the present invention, the luminance has been described as an example of the pixel data, but it is also possible to detect the flushing and the scene change based on other parameters related to the image such as saturation and brightness.

(Second Embodiment) Next, a video feature detecting apparatus according to a second embodiment of the present invention will be described. In the second embodiment, the image feature detecting method or the image feature detecting program for realizing the image feature detecting apparatus described in the first embodiment is recorded in a recording medium such as a floppy (registered trademark) disk. By carrying out the transfer, the processing shown in the above embodiment can be easily carried out in an independent computer system.

FIG. 7 is a diagram for explaining a case where this is implemented with a floppy disk. FIG. 7A is a diagram showing an example of the physical format of a floppy disk, which is the recording medium body. Tracks are formed concentrically from the outer circumference to the inner circumference, and are divided into 16 sectors in the angular direction. There is.

FIG. 7B is a diagram for explaining a case for accommodating this floppy disk. From the left side of the paper, a front view and a sectional view of the floppy disk case and a floppy disk are shown, respectively. By storing the floppy disk in the case in this way, the disk can be protected from dust and external shocks and safely transferred.

FIG. 7C is a diagram for explaining recording and reproducing of a program on the floppy disk. As shown in the figure, by connecting a floppy disk drive to the computer system, it becomes possible to record and reproduce the program on the disk.
The disk is inserted into and removed from the floppy drive via the insertion slot. When recording, the program is recorded from a computer system to a disk by a floppy disk drive. For playback, the floppy disk drive reads the program from disk and transfers it to the computer system.

As described above, the image feature detecting program for realizing the image feature detecting method is recorded in the recording medium such as the floppy disk, so that the first embodiment described above can be realized.
It is possible to easily carry out the processing indicated by.

In the above description, a floppy disk is used as the data recording medium, but an image characteristic detection process by software can be performed using an optical disk as in the case of the floppy disk. Further, the recording medium is not limited to the above-mentioned optical disk or floppy disk, but may be an IC card, a ROM cassette or the like, as long as it can record a program. Similarly, the image feature detection process can be performed.

In the first embodiment described above, the present invention is realized by a device (hardware), but it may be realized by a program (software) executed under a general-purpose CPU. It goes without saying that the same effect as that obtained by hardware can be obtained even by the method by software.

[0044]

As described above, according to the video feature detecting method according to the first aspect of the present invention, the video or the temporal feature of the screen is detected in the screen of the images continuously input in time. And a screen dividing step of dividing the screen of the input image into a plurality of areas.
An area storage step of storing the image data of the divided screen area, an image data of the area stored in the area storage step, and an image of an area temporally continuous with the image data of the stored area are displayed on the screen. Based on the image data obtained by division, a parameter calculation step for calculating a parameter relating to pixel data between these two image data, and a change in the pixel data based on the parameter obtained in the parameter calculation step Since it includes a video screen feature detection step of determining whether or not the time corresponding to the time defined by the threshold group has occurred, it is necessary to correctly detect flushing and scene change with a small amount of data. Since an appropriate motion vector can be obtained based on this detection information, the motion obtained in this way By performing high-compression coding of a moving picture using a vector, there is an advantage that it is possible to degradation obtain an image inconspicuous.

According to a second aspect of the present invention, there is provided the image feature detecting method according to the first aspect, wherein the change of the pixel data is within a time period defined by the first threshold value group. Since it is assumed that the step of determining whether or not a corresponding amount has occurred includes an area-specific determination step of performing determination in each of the divided areas,
The effect that the flushing and the scene change can be distinguished with high accuracy can be obtained.

According to the video feature detecting method of claim 3 of the present invention, in the video feature detecting method of claim 1, the video screen feature detecting step is determined by the first threshold value group. Since the method includes a step of determining whether or not the value calculated from the parameter obtained in the period corresponding to the time corresponds to the value defined by the second threshold value group, it is possible to detect a change in the pixel data. It is possible to prevent the sensitivity from becoming too sensitive, and it is possible to obtain an effect that the flushing or the scene change can be accurately detected.

According to a fourth aspect of the present invention, there is provided the image feature detecting method according to the third aspect, wherein the image feature detecting method is obtained in a period corresponding to the time defined by the first threshold value group. Since the value calculated from the obtained parameter is the difference between the maximum value and the minimum value of the parameter obtained during the period corresponding to the time defined by the first threshold value group, As a result, it is possible to prevent the detection sensitivity from becoming too sensitive, and it is possible to obtain an effect that the flushing or the scene change can be accurately detected.

According to a fifth aspect of the present invention, there is provided the image feature detecting method according to the second aspect, wherein the parameter of each area that is the determination target in the area-specific determination step is A threshold value number detecting step of detecting whether or not the number corresponding to the value defined by the threshold value group of 3 exists as much as the number defined by the fourth threshold value group is included. Therefore, the effect that the scene change and the flushing can be distinguished with high accuracy can be obtained.

According to a sixth aspect of the present invention, there is provided the image feature detecting method according to the fifth aspect, wherein a predetermined number of pixels of the pixel data to be judged are temporally before and after the screen. Regarding the pixel data of the screen of, the value of the parameter of each area obtained by dividing the screen is compared with the value corresponding to the value defined by the third threshold value group,
Since the step of comparing the comparison result with the number determined by the fourth threshold value group is included, it is possible to obtain an effect that the scene change can be detected with high accuracy in distinction from the flushing.

According to a seventh aspect of the present invention, there is provided the image feature detecting method according to the first aspect, in which the calculated parameter is for each pixel in the divided screen area. Since the sum of the signed difference values of the pixel values that change over time is used, it is possible to provide the device with a simple configuration using a general addition / subtraction circuit.

According to an eighth aspect of the present invention, there is provided the image feature detecting method according to the fifth aspect, wherein the sum of the signed difference values between the pixel values is a divided screen area. It is determined by a pixel difference calculation step of calculating a signed difference value over time for each pixel in the area, and an area sum calculation step of calculating the sum within the area of the values calculated in the pixel difference calculation step. So general addition /
The advantage is that the device can be provided in a simple way using a subtraction circuit.

According to a ninth aspect of the present invention, in the image feature detecting method according to the seventh aspect, the sum of the signed difference values between the pixel values is a divided screen area. A pixel sum calculation step for calculating the sum of pixel values within the region, and a region difference calculation step for calculating a signed difference value over time from the values calculated in the pixel sum calculation step,
Since it is determined by the method described above, there is an effect that the device can be provided by a simple method using a general addition / subtraction circuit.

Further, according to the image feature detecting apparatus of the tenth aspect of the present invention, the image feature detecting apparatus for detecting the image or the temporal feature of the screen on the screen of images continuously input in time. And a screen dividing means for dividing the screen of the input video into a plurality of areas, a storage means for storing image data of the divided screen areas, and image data of the areas stored by the storage means. , A parameter relating to pixel data between these two image data is calculated from the image data of the stored region and the image data obtained by dividing the image of the temporally continuous region by the screen dividing means. From the parameter calculation means and the parameter obtained by the parameter calculation means, the change of the pixel data occurs for the time period defined by the first threshold value group. Since it is equipped with a video screen feature detecting means for determining whether or not there is a small amount of data, it is possible to correctly distinguish and detect flushing and scene change. Based on this detection information, an appropriate motion vector can be detected. Therefore, by performing high compression encoding of the moving image using the motion vector thus obtained, it is possible to obtain an image in which deterioration is not noticeable.

According to the eleventh aspect of the data storage medium of the present invention, the image feature for causing the computer to perform the process of detecting the temporal feature of the image or screen which is continuously input in terms of time. A data storage medium storing a detection program, wherein the video feature detection program includes a first program for causing a computer to perform screen division processing for dividing a screen of an input video into a plurality of areas, and the division. A second program for performing a process of storing the image data of the stored screen area by a computer, the image data of the area stored in the second program, and the image data of the stored area that are temporally continuous. From the image data obtained by dividing the image of the area to be divided by the above-mentioned first program, place it between these two image data. From the third program for executing the process of calculating the parameter relating to the pixel data by the computer and the parameter obtained in the third program, the change in the pixel data corresponds to the time defined by the first threshold value group. A computer performs a video screen feature detection process for determining whether or not the occurrence has occurred.
Since the program and is included, the computer can correctly detect the flushing and the scene change with a small amount of data, and an appropriate motion vector can be obtained based on this detection information. By performing high-compression coding of a moving image using the obtained motion vector, it is possible to provide an image feature detection device that can obtain a video in which deterioration is not noticeable.

[Brief description of drawings]

FIG. 1 is a block diagram for explaining a configuration of a video feature detection device according to a first exemplary embodiment of the present invention.

FIG. 2 is a diagram showing a flowchart showing a flow of processing performed by the video feature detection apparatus according to the first embodiment.

FIG. 3 is a schematic diagram showing an example of screen division by a screen division unit which constitutes the video feature detection apparatus.

FIG. 4 is a schematic diagram for explaining an example of obtaining a difference between adjacent field screens using the above-mentioned video feature detection device.

FIG. 5 is a schematic diagram for explaining a temporal transition of a sum of signed difference values of pixel values of one block divided by the screen dividing unit and a flushing detection procedure.

FIG. 6 is a schematic diagram for explaining a scene change detection procedure.

FIG. 7 is a schematic diagram for explaining a program recording medium for storing a program for implementing a video feature detection method by a computer system according to a second embodiment of the present invention.

[Explanation of symbols]

101 screen division means 102 area storage means 103 parameter calculation means 104 video screen feature detection means

Continued front page    F term (reference) 5B057 CC02 CE09 CH01 CH11 DA08                       DC01                 5C059 KK01 MA00 NN43 PP04 SS20                       TA21 TA66 TB06 TC02 TC14                       TD05 TD12 UA38                 5C076 AA36 BA06                 5L096 GA19 HA04 JA11

Claims (11)

[Claims] 1. A video feature detection method for detecting temporal features of a video or a screen in a screen of video that is continuously input in time, wherein the screen of the input video is divided into a plurality of areas. Screen division step, an area storage step of storing image data of the divided screen area, image data of the area stored in the area storage step, and the image data of the stored area are temporally consecutive. From the image data obtained by dividing the image of the area to be divided into screens,
From the parameter calculation step of calculating the parameter relating to the pixel data between these two image data and the parameter obtained in the parameter calculation step, the change of the pixel data corresponds to the time defined by the first threshold value group. A video screen feature detection step of determining whether or not the video feature has occurred. 2. The image feature detecting method according to claim 1, wherein the step of determining whether or not the change of the pixel data has occurred for a time period defined by the first threshold value group, An image feature detection method comprising a region-by-region determination step of performing determination in each of the divided regions. 3. The video feature detecting method according to claim 1, wherein the video screen feature detecting step includes a step of calculating a value calculated from a parameter obtained during a period corresponding to the time defined by the first threshold value group. An image feature detecting method, comprising: a step of determining whether or not the value corresponds to a value defined by a second threshold value group. 4. The image feature detecting method according to claim 3, wherein the value calculated from the parameter obtained during the period corresponding to the time defined by the first threshold value group is the first threshold value. A video feature detection method, which is a difference between a maximum value and a minimum value of a parameter obtained in a period corresponding to a time defined by a value group. 5. The image feature detecting method according to claim 2, wherein the number of parameters of each region that is the determination target in the region-specific determination step corresponds to a value defined by a third threshold value group, A video feature detection method comprising a threshold number detection step of detecting whether or not there are as many as the number defined by the fourth threshold group. 6. The image feature detecting method according to claim 5, wherein the pixel data of a predetermined number of screens temporally before and after the screen of the pixel data as the determination target are obtained by dividing the screen. Comparing the value of the parameter of the region with a value corresponding to the value defined by the third threshold group, and comparing the comparison result with the number defined by the fourth threshold group. A method for detecting a characteristic image feature. 7. The image feature detection method according to claim 1, wherein the calculated parameter is a sum of signed difference values of pixel values that change over time for each pixel in the divided screen area. A video feature detection method characterized by: 8. The image feature detection method according to claim 5, wherein the sum of the signed difference values between the pixel values is a pixel for calculating a signed difference value over time for each pixel in the divided screen area. A video feature detection method comprising: a difference calculation step; and a region sum calculation step of calculating a sum of the values calculated in the pixel difference calculation step within a region. 9. The image feature detection method according to claim 7, wherein the sum of the signed difference values between the pixel values is a pixel sum calculation step of calculating a sum of pixel values in the divided screen areas, An area difference calculation step of calculating a time-coded difference value from the value calculated in the pixel sum calculation step, and a video characteristic detection method. 10. A video feature detecting apparatus for detecting temporal features of a video or a screen in a screen of video continuously input in time, wherein the screen of the input video is divided into a plurality of areas. Screen dividing means, storage means for storing the image data of the divided screen area, image data of the area stored by the storage means, and an area temporally continuous with the image data of the stored area From the image data obtained by dividing the image of the image by the screen dividing means, and a parameter calculating means for calculating a parameter relating to pixel data between these two image data, and a parameter obtained by the parameter calculating means, A video screen feature detecting means for determining whether or not the change in the pixel data has occurred for a time period defined by the first threshold value group. Image feature detector, characterized in that. 11. A data storage medium storing a video feature detection program for causing a computer to perform a process of detecting temporal features of images or screens that are continuously input in terms of time. The detection program uses a computer to execute a first program for causing a computer to perform a screen division process for dividing an input video screen into a plurality of regions, and a process for storing the image data of the divided screen region. The second program to be executed, the image data of the area stored in the second program, and the image of the area temporally continuous with the image data of the stored area are screen-divided by the first program. From the obtained image data, the process of calculating the parameter relating to the pixel data between these two image data is It is determined from the third program executed by the computer and the parameters obtained in the third program whether or not the change in the pixel data has occurred for the time defined by the first threshold value group. A data storage medium comprising: a fourth program for performing a video screen feature detection process by a computer.