JP2007306450A - Cinema decision method, cinema decision apparatus, and caption detecting method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To make a cinema decision with high accuracy even for an input picture containing a picture and a caption. <P>SOLUTION: A frame difference calculator 101 calculates the frame difference value between an input signal in the present filed of an input picture and an input signal previous to the present field every pixel on the screen, an entire frame difference accumulator 102 accumulates their frame difference values, a character detector 103 detects characters to determine a caption region window by a character region detector 104, a character region frame difference accumulator 105 calculates the frame difference values of the input signals for all pixels in the caption region window determined by the detector 104 and accumulates the frame difference values of all the pixels in the window, and a frame difference subtracter 106 subtracts the accumulated frame difference values from the frame difference values accumulated by the accumulator 102 to output the result. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a cinema determination method for determining whether an input video is a cinema source, a cinema determination device, and a subtitle detection method for an input video.

  A movie source, which is a progressive signal of 24 frames (or 25 frames), is converted into a 60-field (or 50-field) interlace signal by performing a static inset process according to the regularity of pull-down, and faithful video Reproducing is called digital cinema reality. In order to realize this, it is necessary to determine in advance whether or not the input video is a cinema source (hereinafter, this determination is referred to as cinema determination).

In a conventional cinema determination apparatus that performs cinema determination, the correlation value of each image in the first and second fields of the input video is calculated, and based on the correlation value, it is determined whether or not it is a duplicate field. There is one that determines whether or not (see, for example, Patent Document 1).
JP-A-8-289199

  However, characters (captions) may be added to the cinema source, and the added characters do not include the pull-down regularity. Therefore, if cinema detection is performed with both video and subtitles included, There is a possibility of misjudging whether it is a source.

  The present invention has been made paying attention to the above problem, and an object of the present invention is to enable cinema determination with high accuracy even when an input video includes both video and subtitles.

In order to solve the above problems, the first aspect of the present invention is:
A cinema determination method for determining whether an input video is a cinema source,
A frame difference calculation step for calculating, for each pixel on the screen, a frame difference value between the input signal of the current field of the input video and the input signal before the current field;
A frame difference value calculated in the frame difference calculation step, a full screen frame difference accumulation step for accumulating the entire screen for the entire screen;
A character detection step of scanning the inside of a caption detection area, which is a predetermined area on the screen, and detecting characters displayed in the area;
Based on the character detection result in the character detection step, a character area detection step for determining a caption area window that is an area in which the caption is displayed by determining a start point and an end point of the caption.
A character that calculates the frame difference value of the input signal for all the pixels in the caption area window determined by the character area detection step and accumulates the frame difference values for all the pixels in the window. Region frame difference accumulation step;
A frame difference subtraction step for subtracting and outputting the frame difference value accumulated in the character region frame difference accumulation step from the frame difference value accumulated in the full screen frame difference accumulation step;
Based on the output of the frame difference subtracting step, the regularity of the pull-down of the input video is detected in field time units, and according to the detection result, the input image is converted and screen output is performed by the cinema determination step,
It is provided with.

  Thereby, a caption area window which is an area where captions are displayed is determined, and cinema determination is performed using an area excluding the caption area window. Therefore, even if the input video includes both video and subtitles, the cinema can be determined with high accuracy.

In addition, the second aspect of the present invention includes
Determining a subtitle detection area for character scanning on the screen;
Scanning the subtitle detection area to detect a character, and determining a start point of the subtitle area window based on the character detection point;
Detecting in which direction the characters have continuity from the start point of the caption area window;
Performing character scanning to determine the end point of the subtitle area window;
It is characterized by having.

  Thereby, the position where the subtitle is displayed on the screen is detected.

The third aspect of the present invention is as follows.
A cinema determination device for determining whether an input video is a cinema source,
A frame difference calculation unit that calculates the frame difference value of the input signal of the current field of the input video and the input signal before the current field for each pixel on the screen;
A frame difference value calculated by the frame difference calculation unit, and a full screen frame difference accumulation unit that accumulates the entire screen;
A character detection unit that scans the subtitle detection area, which is a predetermined area on the screen, and detects characters displayed in the area;
Based on the character detection result in the character detection unit, a character area detection unit that determines a subtitle area window that is an area in which the subtitle is displayed by determining a start point and an end point of the subtitle;
A character that calculates the frame difference value of the input signal for all the pixels in the caption area window determined by the character area detection unit and accumulates the frame difference values for all the pixels in the window. An area frame difference accumulation unit;
A frame difference subtraction unit that subtracts and outputs the frame difference value accumulated in the character region frame difference accumulation unit from the frame difference value accumulated in the full-screen frame difference accumulation unit;
Based on the output of the frame difference subtracting unit, the regularity of the pull-down of the input video is detected in field time units, and according to the detection result, the cinema determination unit that converts the input video and outputs the screen,
It is provided with.

  Thereby, a caption area window which is an area where captions are displayed is determined, and cinema determination is performed using an area excluding the caption area window. Therefore, even if the input video includes both video and subtitles, the cinema can be determined with high accuracy.

  According to the present invention, the subtitle area window, which is the area where the subtitle is displayed, is determined, and the cinema determination is performed using the area excluding the subtitle area window, so that the input video includes both the video and the subtitle. Even if there is, the cinema can be judged with high accuracy.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram showing a configuration of a cinema determination device 100 according to an embodiment of the present invention. As shown in the figure, the cinema determination apparatus 100 includes a frame difference calculation unit 101, a full screen frame difference accumulation unit 102, a character detection unit 103, a character area detection unit 104, a character area frame difference accumulation unit 105, and a frame difference subtraction unit 106. , And a cinema determination unit 107.

  The frame difference calculation unit 101 calculates a frame difference value between the input signal (input luminance signal) in the current field of the input video and the input signal two fields or one field before for each pixel on the screen.

  The full screen frame difference accumulation unit 102 accumulates the frame difference values calculated by the frame difference calculation unit 101 for the entire screen.

  The character detection unit 103 scans the inside of a predetermined caption detection area (described later) on the screen and detects characters displayed in this area. Specifically, the character detection unit 103 detects a pattern in which the video signal level is lower than the black level threshold level, then higher than the white level threshold level, and lower than the black level threshold level. Are recognized as white characters. FIG. 2 is an example showing a signal level change pattern when subtitles are displayed at the bottom of the screen. Here, the reason why the threshold value is provided for the determination of white and black is to prevent the erroneous detection of characters by giving the black level / white level a certain continuous width. In the present embodiment, as the caption detection area, areas on the top, bottom, left and right (candidate areas) of a screen on which a caption appears are set (see FIG. 3). By setting the caption detection area in this way, it is possible to prevent caption error detection.

  Based on the character detection result of the character detection unit 103, the character region detection unit 104 determines the start point and the end point of the subtitle, and determines the region where the subtitle is displayed (subtitle region window). Yes. Specifically, the character area detection unit 104 determines the caption area window as follows.

  First, the character detection unit 103 detects characters by sequentially scanning from the scanning start point set in advance in the caption detection area. For example, in the example shown in FIG. 4, the scan start point is for generating the upper horizontal window on the screen, generating the vertical window on the right side of the screen, generating the vertical window on the left side of the screen, and generating the horizontal window on the lower part of the screen Four points are determined as start points, and scanning is sequentially performed for each start point in a direction (direction of an arrow in FIG. 4) determined for each start point. As shown in FIG. 4, a horizontal window is generated by scanning in the horizontal direction, or a vertical window in which characters are vertically generated is generated. There are two cases (see FIG. 5).

  The upper left part from the pixel point (character detection point) where the character is first detected is determined as the start point of the caption area window. At this time, if the character detection point is determined as the start point of the caption area window as it is, when the beginning of the appearance character is “A” or “C”, the subtitle area window cannot enclose the entire subtitle. A start point is determined by taking a margin in the left direction and the upward direction, and a caption area window is created wider.

  When the start point of the caption area window is determined, it is detected in which direction the character has continuity next. For this purpose, the character detection unit 103 is caused to scan characters in the vertical direction or the horizontal direction from the start point of the caption area window. At this time, in order to see the continuity of the character, a threshold value is provided for determining an interval from the time when one character is detected until the next character is detected. This threshold value is provided in any case where the scanning direction is vertical or horizontal.

  After the direction in which the characters continue is determined, the characters are continuously scanned to determine the end point of the caption. If no character is detected even after scanning for a predetermined interval, that point is detected as the end point of the caption detection window. The end point of the caption detection window is also set in the lower right direction with a margin between the setting values in the right direction and the lower direction.

  On the other hand, the character area frame difference accumulating unit 105 calculates the frame difference value of the input signal for all the pixels in the subtitle area window determined by the character area detecting unit 104, and all the pixels in the window. The frame difference values for these pixels are accumulated.

  The frame difference subtracting unit 106 subtracts the frame difference value accumulated by the character area frame difference accumulating unit 105 from the frame difference value accumulated by the full-screen frame difference accumulating unit 102, and the subtraction result is determined in the subsequent cinema determination. This is sent to the unit 107. That is, the output of the frame difference subtraction unit 106 is a cumulative value of frame differences only for the video portion that does not include subtitles.

  The cinema determination unit 107 performs screen output as follows. That is, the cinema determination unit 107 detects the regularity of the pull-down of the input video in field time units based on the output of the frame difference subtraction unit 106, and if the input video is determined to be a cinema source, the full screen area For all the pixels except the subtitle area, the input source is statically fitted and output to the screen. In addition, regarding the pixels in the subtitle area window for the time when the input video is not determined to be a cinema source, the time for entering the cinema source detection sequence, and the time when the input is completely determined as a cinema source, the input video A normal IP (Interlace-Progressive) conversion for performing moving image still mixing based on the information of the motion detection result is performed, and screen output is performed.

  In the cinema determination apparatus 100 described above, first, the frame difference calculation unit 101 displays the frame difference value of the input signal (input luminance signal) of the current field of the input video and the input signal two fields (or one field) on the screen. The frame difference value calculated by the frame difference calculation unit 101 is accumulated for the entire screen by the full screen frame difference accumulation unit 102.

  On the other hand, the character region detection unit 104 determines the start point and end point of the caption based on the character detection result of the character detection unit 103, and determines the region (caption region window) where the caption is displayed. Then, the character area frame difference accumulating unit 105 calculates the frame difference value of the input signal for all the pixels in the caption area window determined by the character area detecting unit 104, and is further in this window. The frame difference values for all pixels are accumulated.

  Next, the frame difference subtracting unit 106 subtracts the frame difference value accumulated by the character region frame difference accumulating unit 105 from the frame difference value accumulated by the full-screen frame difference accumulating unit 102, and the subtraction result is shown in the subsequent stage. To the cinema determination unit 107. Thereafter, the cinema determination unit 107 performs cinema determination based on the subtraction result of the frame difference subtraction unit 106, converts the input video according to the determination result, and performs screen output.

  As described above, according to the present embodiment, the subtitle area is detected by comparing the input signal level with the white and black threshold levels to detect characters, and cinema determination is performed using the area excluding the subtitle area. Therefore, even if the input video includes both video and subtitles, the cinema can be determined with high accuracy.

  Since the cinema determination method, the cinema determination device, and the caption detection method according to the present invention determine a caption area window that is an area where captions are displayed, and perform cinema determination using an area excluding the caption area window. Cinema determination method, cinema determination device, and input video subtitle for determining whether input video is a cinema source, having the effect of enabling high-precision cinema determination even when the input video includes both video and subtitles This is useful as a detection method.

It is a block diagram which shows the structure of the cinema determination apparatus which concerns on embodiment of this invention. It is a figure which shows the signal level change pattern in case a subtitle is displayed on the screen lower part. It is a figure which shows the example of a setting of a subtitle detection area. It is a figure which shows an example of a scanning start point and a scanning direction. It is a figure which shows the example in which the horizontal window in which a character continues horizontally, and the example in which the vertical window in which a character continues vertically are generated.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Cinema determination apparatus 101 Frame difference calculation part 102 Full screen frame difference accumulation part 103 Character detection part 104 Character area detection part 105 Character area frame difference accumulation part 106 Frame difference subtraction part 107 Cinema determination part

Claims (3)

A cinema determination method for determining whether an input video is a cinema source,
A frame difference calculation step for calculating, for each pixel on the screen, a frame difference value between the input signal of the current field of the input video and the input signal before the current field;
A frame difference value calculated in the frame difference calculation step, a full screen frame difference accumulation step for accumulating the entire screen for the entire screen;
A character detection step of scanning the inside of a caption detection area, which is a predetermined area on the screen, and detecting characters displayed in the area;
Based on the character detection result in the character detection step, a character area detection step for determining a caption area window that is an area in which the caption is displayed by determining a start point and an end point of the caption.
A character that calculates the frame difference value of the input signal for all the pixels in the caption area window determined by the character area detection step and accumulates the frame difference values for all the pixels in the window. Region frame difference accumulation step;
A frame difference subtraction step for subtracting and outputting the frame difference value accumulated in the character region frame difference accumulation step from the frame difference value accumulated in the full screen frame difference accumulation step;
Based on the output of the frame difference subtracting step, the regularity of the pull-down of the input video is detected in field time units, and according to the detection result, the input image is converted and screen output is performed by the cinema determination step,
A cinema judgment method characterized by comprising: Determining a subtitle detection area for character scanning on the screen;
Scanning the subtitle detection area to detect a character, and determining a start point of the subtitle area window based on the character detection point;
Detecting in which direction the characters have continuity from the start point of the caption area window;
Performing character scanning to determine the end point of the subtitle area window;
A cinema subtitle detection method characterized by comprising: A cinema determination device for determining whether an input video is a cinema source,
A frame difference calculation unit that calculates the frame difference value of the input signal of the current field of the input video and the input signal before the current field for each pixel on the screen;
A frame difference value calculated by the frame difference calculation unit, and a full screen frame difference accumulation unit that accumulates the entire screen;
A character detection unit that scans the inside of a caption detection area, which is a predetermined area on the screen, and detects characters displayed in the area;
Based on the character detection result in the character detection unit, a character area detection unit that determines a subtitle area window that is an area in which the subtitle is displayed by determining a start point and an end point of the subtitle;
A character that calculates the frame difference value of the input signal for all the pixels in the caption area window determined by the character area detection unit and accumulates the frame difference values for all the pixels in the window. An area frame difference accumulation unit;
A frame difference subtraction unit that subtracts and outputs the frame difference value accumulated in the character region frame difference accumulation unit from the frame difference value accumulated in the full-screen frame difference accumulation unit;
Based on the output of the frame difference subtracting unit, the regularity of the pull-down of the input video is detected in field time units, and according to the detection result, the cinema determination unit that converts the input video and outputs the screen,
A cinema determination device characterized by comprising:

Images