KR100311739B1 - Method of cuts-detection of moving picture - Google Patents

Abstract

The present invention provides a new video scene transformation detection method modified by combining the advantages of a color histogram and an X ² scene transformation detection method among algorithms for calculating a histogram difference value in order to construct a database from a video. There is a purpose.

According to the present invention, in the moving picture scene detection method for detecting a converted video between successive moving picture frames, R (Red) histogram, G (Green) histogram, and B (Blue) histogram of the previous video to the next video are obtained. The first step; A second step of obtaining a histogram difference value of the next video continuous to the previous video using the R histogram to the B histogram; And comparing the histogram difference value with a predetermined value determined by an operator, thereby detecting a scene converted between successive video frames.

As a result, it is possible to detect a scene converted between moving images through a simpler calculation.

[Index]

Video scene change detection method, histogram, video scene, weight, predetermined value.

Description

{Method of cuts-detection of moving picture}

The present invention relates to a scene transformation detection method for constructing a database from a video. In particular, after storing each frame of an input video in an image file format, an RGB histogram of each image is obtained to obtain a histogram. A method for detecting a scene transformation using a comparison method.

Recently, as the utilization of multimedia data is actively demanded, a database technology for storing, managing and retrieving various data is required. In particular, video, which is a video material, provides a lot of information to human beings very naturally, and if there is a lot of interest, researches on the construction of a content-based multimedia database are actively conducted to store and use a database for a specific problem area. In the process.

Meanwhile, in order to efficiently use a multimedia database, a scene segmentation method of a video is required. In the related art, a method of using a difference between frames, a method of extracting and using an edge, and a whole wavelet are measured and used. There are many ways, such as the method. Especially, the method of calculating and using a histogram difference is common.

However, the conventional scene segmentation method that calculates the histogram difference and divides the scene is very sensitive to the movement or noise of the object to produce an unsatisfactory result. There is a disadvantage, and there is a problem that is difficult even when applying a predetermined value.

The present invention has been made to solve the problems of the prior art as described above, among the conventional algorithms for calculating the histogram difference value, the advantages of the color histogram and the X ² scene transformation detection method to simplify the calculation While doing so, it is an object of the present invention to provide a method for detecting scene transformation of a video with precision and reliability.

1 is an overall configuration diagram of a video analysis system for explaining a general video scene transformation detection method;

2 is a flowchart illustrating a scene change detection method of a video according to an embodiment of the present invention;

3 is a flowchart illustrating a process of calculating a histogram difference value in a video scene transformation detection method according to the present invention;

4 is a view comparing a histogram difference value according to an embodiment of the present invention with results according to another conventional method;

5 is a view showing the results for various materials according to an embodiment of the present invention,

FIG. 6 is a diagram illustrating a predetermined value applied to a histogram difference value used in a video image transition detection method according to the present invention; FIG.

7 is a graph illustrating a result of detecting a video scene change according to the present invention and a result according to another conventional method.

According to the present invention for achieving the object as described above, in the video scene conversion detection method for detecting the converted video between successive video frames, each R (Red) histogram, G ( Green) a first step comprising a histogram and a B (Blue) histogram; A second step of obtaining a histogram difference value of a next video continuous to a previous video using the R histogram to the B histogram; And comparing the histogram difference value with a predetermined value determined by an operator, thereby detecting a scene converted between successive video frames.

Preferably, in a recording medium storing a computer-executable program for performing a video scene conversion detection method through an RGB histogram and an X ² histogram, each R (Red) histogram, G of a previous video to the next video, G A first step of obtaining (Green) histogram and B (Blue) histogram; A second step of obtaining a histogram difference value of a next video continuous to a previous video using the R histogram to the B histogram; And comparing the histogram difference value with a predetermined value determined by an operator, thereby detecting a scene converted between successive video frames. Is provided.

Hereinafter, a video scene transformation detection method according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings. 1 is an overall configuration diagram of a video analysis system for explaining a general video scene transformation detection method.

The video 10 stored in the video or video tape directly input to the video camera is stored as a video file 20 using a capture device. After converting and storing each frame of the stored video file 20 into the still image file 30, the histogram difference value of the stored still image file is calculated to detect a scene change (extract cut frame 40). A key frame is selected 50 from the cut frames. When the video indexing 60 is configured using the selected key frame, the user 80 may search for the video file stored in the video file format using the search browser 70.

2 is a flowchart illustrating a video scene change detection method according to an exemplary embodiment of the present invention.

First, in step S201 to step S204, each histogram of the RGB of each file is obtained from the first video file to the last video file, and then in step S205, a difference value between the previous video and the next video is calculated, and the difference value is calculated. It is judged whether or not it is larger than the predetermined value designated by the operator. If the difference value is larger than the predetermined value, it is selected in step S206 as the cut frame, and if it is smaller or equal, the process returns to step S202 via step S207. After that, the same procedure is repeated until the last image.

3 is a flowchart illustrating a histogram difference value calculation according to an embodiment of the present invention.

First, in steps S301 to S303, the values corresponding to the red of the histogram of the current image and the next image are subtracted from the elements 0 to 225 of each of the histograms of the image, and the value is squared.

Subsequently, in step S304, the squared value is divided by the histogram element value of the next image for shaping, and then the value is multiplied by the weight according to the National Television Systen Committee (NTSC) standard.

In the same manner, in Steps S305 to S308, the values for the Green and Blue elements are obtained, and in Step S309, the difference values of the respective RGB are added, and then divided by 3 to accumulate. The above method is repeated in step S310 up to 255, which is the end element of the histogram.

When the method is expressed in the form of an equation, it is expressed as Equation 1 below.

[Equation 1]

[Equation 1] is described in detail as follows.

d (I _i , I _j ) means the difference between the i th image and the j th image, Denotes the value of the k th element in the histogram of the red of the i th image. That is, after subtracting the same elements of the histogram for the red of the previous image and the histogram for the red of the next image, the value is squared to emphasize the difference. The squared value is divided by the histogram value of the next image for shaping. In this way, the values for Green and Blue are calculated, and the values are accumulated by multiplying the values according to the NTSC standard.

On the other hand, the weights to be used in the present embodiment are 0.229, 0.587 and 0.114, respectively, which are used when converting from RGB color to intensity class, and used to convert the RGB color model to the Y component of the YIQ model. This value is one of the standard methods specified by NTSC and is a standard method that can be used when converting to another model.

For the experiment of this embodiment, the histogram difference value was calculated using scene data as shown in Table 1 below to detect the scene transformation.

TABLE 1

FIG. 4 is a diagram comparing a histogram difference value according to an embodiment of the present invention with a result according to another conventional method, and is a histogram difference value calculated by using A among video data of Table 1. FIG.

Among the existing methods, the color histogram method 41 and the bin_to_bin difference method 43 are very complicated in histogram, so that the result of scene change detection according to a predetermined value is very sensitive. In addition, it is difficult to set an appropriate predetermined value, and there is a disadvantage that the resultant value varies greatly even with a small change of the predetermined value.

However, the use of the X ² histogram method 42 and the method 44 according to an embodiment of the present invention has the advantage that the scene transformation is located mostly at the place where the histogram difference is large, so that a predetermined value selection can be easily made.

In particular, the method 44 according to an embodiment of the present invention, if the histogram difference value is not large enough, only the scenes where the difference in the value is large appear in the histogram, so that the predetermined value can be automatically set. Provide the foundation.

5 is a graph showing the results of applying the method according to an embodiment of the present invention to the data in [Table 1]. , Results plot applied to data C (430), results plot applied to data D (440), results plot applied to data E (450), results plot applied to data F (460), results plot applied to data G (470) And Result 480 applied to Data H. FIG.

FIG. 6 is a graph in which a predetermined value is applied to a histogram difference value according to an embodiment of the present invention, and a predetermined value is applied by applying a predetermined value 472 to a result diagram of the histogram difference value applied to the data G of [Table 1]. The image indicated by the higher value 471 is determined as the scene change point, and the image is selected as the cut frame.

7 is a graph comparing a result according to an embodiment of the present invention with a result according to another conventional method, and a bar means a case in which an error frame that should not be extracted is extracted, and according to an embodiment of the present invention. It can be seen that the results are much more accurate than the other methods.

As described in detail above, the present invention combines the advantages of the color histogram X ² scene transform detection method among the algorithms for calculating the histogram difference value in the scene image detection method while performing a simpler calculation. This has the effect of being accurate and reliable.

Although the technical spirit of the present invention has been described above with the accompanying drawings, this is for illustratively describing the best embodiments of the present invention and not for limiting the present invention. In addition, it is obvious that any person skilled in the art may make various modifications and imitations without departing from the scope of the technical idea of the present invention.

Claims (6)

A moving picture scene detection method for detecting a converted video between successive moving picture frames, comprising: a first step of obtaining R (Red) histograms, G (Green) histograms, and B (Blue) histograms of a previous video or a next video; Step; A second step of obtaining a histogram difference value of a next video continuous to a previous video using the R histogram to the B histogram; And comparing the histogram difference with a predetermined value determined by an operator, thereby detecting a scene converted between successive video frames. The method of claim 1, wherein the histogram difference value is generated by providing values of the R, G, and B histogram values of the previous image and the next image except for the same element. The video scene change detection method according to claim 1 or 2, wherein if the difference value is larger than the predetermined value, the video frame is set as a cut frame. The method of claim 1 or 2, wherein the second step comprises: a first sub-step for shaping and dividing the histogram difference value by each histogram element value of the next image; A second substep of multiplying the weighted value by the corresponding histogram value; A third substep of adding all the difference values of the R histogram to the B histogram value obtained through the second substep, dividing by 3, and accumulating; And a fourth substep of repeating the first to third substeps to the end elements of the R histogram and the B histogram. 5. The method according to claim 4, wherein the weight is in accordance with the standard of the National Television System Committee (NTSC). There is a recording medium that stores a computer-executable program for performing a video scene conversion detection method using an RGB (Red, Green, Blue) histogram and an X ² histogram, each R (Red) histogram of the previous or next video, A first step having a G (Green) histogram and a B (Blue) histogram; A second step of obtaining a histogram difference value of a next video continuous to a previous video using the R histogram to the B histogram; And comparing the histogram difference value with a predetermined value determined by an operator, thereby detecting a scene converted between the successive moving image frames.