JP3609236B2 - Video telop detection method and apparatus - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a video telop detection method and apparatus for detecting a telop from a video.
[0002]
[Prior art]
As the distribution of images such as TV broadcasting, digital satellite broadcasting, laser discs, DVDs, and Video on Demand increases, the demand for more flexible handling of images has increased. Therefore, a technique has been proposed in which various table of contents and index information are added to a video to enable video search and random access. Of the information that characterizes the video, subtitles ,symbol The telop such as is important as reflecting the meaning and content of the video. Therefore, a method for automatically detecting a telop from a video has been proposed.
[0003]
Conventional video telop detection devices include the following.
[0004]
For example, there is an apparatus that uses a property that a telop is stationary and has high accuracy to obtain a portion that does not move between two frames, and further detects a region having a luminance of a predetermined value or more as a subtitle portion. (Japanese Patent Laid-Open No. 8-331456 “Subtitle Moving Device”).
[0005]
Also, there is a method of detecting a block in which the edge strength and luminance of a frame image exceed a threshold value as a portion where a caption exists, using the property that a telop has a strong edge and high accuracy (for example, Japanese Patent Laid-Open No. Hei 8). -21231 “Subtitle detection method and moving image representative image extraction device”).
[0006]
Also, in a video encoded using inter-frame correlation such as MPEG, pixels encoded using inter-frame correlation and without motion compensation are temporally and spatially concentrated on the telop portion. It has the nature of There is also a device that uses this property to detect a telop by counting the frequency of appearance of pixels encoded using correlation between frames and without using motion compensation in a certain time interval (1996 Electronic Information and Communication Society Information and System Society Conference "Telop area detection method from MPEG video").
[0007]
[Problems to be solved by the invention]
In the conventional technology described above, telops are detected using the properties of telops such as edge strength and luminance, but the ability to detect telop switching is low. For example, in Japanese Patent Laid-Open No. 8-212231, a frame image is divided into blocks, and caption area data corresponding to the blocks is prepared. In the caption area data, 1 is stored in a block where a telop is present and 0 is stored in a block where the telop is not present. Then, the number of blocks having different subtitle area data values between the two frames is counted, and the telop is switched when the number exceeds a certain ratio.
[0008]
However, in this method, when the telop is switched without any change and the area is not changed, no change appears in the caption area data, so that the telop switching cannot be detected.
[0009]
An object of the present invention is to provide a video telop detection method and apparatus for detecting telop switching from video.
[0010]
[Means for Solving the Problems]
The video telop detection method of the present invention includes a telop candidate image creation stage for creating a telop candidate image indicating an area where a telop candidate exists from a frame image, a difference image creation stage for creating a difference image of two frame images, An extraction stage for extracting only a portion where a telop candidate exists from the difference image and a determination stage for determining the presence of the telop based on the telop candidate image and the extracted difference image are included.
[0011]
The video telop detection device of the present invention also includes a telop candidate image creating unit that creates a telop candidate image indicating a region where a telop candidate exists from a frame image, and a difference image creating unit that creates a difference image between two frame images. And extraction means for extracting only a portion where a telop candidate exists from the difference image, and determination means for determining the presence of the telop based on the telop candidate image and the extracted difference image.
[0012]
The telop candidate image creating means detects a telop candidate from the frame image and creates a telop candidate image. The difference image creating means creates two frame images. The extraction means extracts a difference image in an area where telop candidates exist. The determination means determines the presence of a telop using a telop candidate image and a difference image of a region where the telop candidate exists. Thereby, since the telop is detected based on the change of the frame image in the area where the telop exists, the change of the telop can be detected.
[0013]
Another video telop detection method of the present invention is:
A telop candidate image creating stage for creating a telop candidate image indicating an area where a telop candidate exists from a frame image;
A difference image creation stage for creating a difference image of two frame images;
Using the telop candidate image and the difference image, a mask creating step of creating a mask indicating a region in which a change in pixel value is to be detected;
Extracting only the portion of the mask from the difference image;
Determining the presence of a telop based on the telop candidate image and the extracted difference image.
[0014]
Further, another video telop detection device of the present invention includes a telop candidate image creating unit that creates a telop candidate image indicating a region where a telop candidate exists from a frame image, and a difference image that creates a difference image between two frame images. Creation means, mask creation means for creating a mask indicating an area in which a change in pixel value should be detected using the telop candidate image and the difference image, extraction means for extracting only the mask portion from the difference image, and telop candidate A determination unit configured to determine presence of a telop based on the image and the extracted difference image;
[0015]
In the conventional technique, basically, the presence / absence of a telop has been determined. Therefore, in the present invention, the change of the telop is detected by examining the change of the frame image in the area where the telop exists.
[0016]
The mask creation image creates a mask using the telop candidate image and the difference image. In the mask, an area where a change in pixel value is to be detected is set to 1, for example, and other areas are set to 0, for example. The extraction means extracts a difference image of the area where the mask is 1, and the determination means determines the telop material using the telop candidate image and the difference image of the area where the mask is 1. As a result, the region for checking the change of the frame image can be limited to the region where the telop exists, so that it is less affected by the change in the background of the telop and the change of the telop can be detected more accurately.
[0019]
According to the embodiment of the present invention, the telop candidate image creating unit extracts the frame that is encoded using the correlation between the frames from the encoded video data, and the macroblock of the extracted frame Counting means for counting the number of macroblocks encoded using correlation between frames without using motion compensation within a fixed counting interval for each macroblock position, and a counting result And a selection means for outputting a matrix representing a telop area by comparing with a given threshold value, and a connected area creating means for obtaining a connected area of the matrix and outputting each connected area as a telop candidate image.
[0020]
According to the embodiment of the present invention, the mask creating means detects an invariant pixel detecting means for detecting an unnecessary pixel which is a pixel having a small value between frames, a new candidate pixel detecting means for detecting a newly appearing telop candidate, Mask composing means for composing a mask indicating an area where at least one of an invariant pixel or a new candidate pixel exists.
[0021]
In the mask creating means, the invariant pixel detecting means detects an area where the pixel is unchanged. The new candidate pixel detection means detects a newly appearing telop candidate. The synthesizing unit synthesizes a mask by obtaining an area where at least one of an invariant pixel and a new candidate pixel exists. As a result, a mask is created in which a pixel in which a telop has existed before and a pixel in which a new telop candidate appears is set to 1, for example, and the other area is set to 0, for example.
[0022]
According to the embodiment of the present invention, the new candidate pixel detecting means includes a first buffer for storing one frame of telop candidate images, a telop candidate image for the current frame, and telop candidate images stored in the first buffer. A first comparison unit that outputs a telop candidate image of the current frame when the telop candidate image exists only in the current frame, and the invariant pixel detection unit includes a second comparison unit that compares the difference image with a threshold value; When the difference image is larger than the threshold value, the second buffer for storing the current frame telop candidate image for one frame and the current frame telop candidate image output from the first comparison means or the second buffer are masked. Mask output means for outputting as an output is included.
[0028]
DETAILED DESCRIPTION OF THE INVENTION
Next, embodiments of the present invention will be described with reference to the drawings.
[0029]
FIG. 1 is a block diagram of a video telop detection apparatus according to a first embodiment of the present invention.
[0030]
This video telop detection apparatus includes a video input terminal 101, a telop candidate image creation unit 102, a difference image creation unit 103, an AND circuit 104, a determination unit 105, and a telop output terminal 106. The telop candidate image creation unit 102 detects a telop from the video input to the video input terminal 101, and creates a telop candidate image in which the pixel where the telop exists is “1” and the other pixels are “0”. The difference image creating unit 103 creates a difference image having an absolute value of a difference between two frame images separated in time as a pixel value. The frame interval may be between adjacent frames or may be an interval separated by a plurality of frames. The AND circuit 104 calculates the product of the telop candidate image and the difference image, and sets the portion of the difference image where the telop candidate image is “0” to “0”. The determination unit 105 determines the presence of a telop using the telop candidate image output from the telop candidate image creation unit 102 and the difference image output from the AND circuit 104, and outputs the telop to the telop output terminal 106. Note that a pixel in which a telop is present may be “0”, other pixels may be “1”, and values other than “0” and “1” may be taken.
[0031]
The advantage of this embodiment is that the difference image is evaluated only in the portion where the telop candidate image exists, so that the telop switching can be detected.
[0032]
Note that conventional techniques can be used to detect telop candidate images. For example, a method based on the above-mentioned 1996 Information and System Society Conference of the Institute of Electronics, Information and Communication Engineers “Telop area detection method from MPEG video” can be used.
[0033]
FIG. 2 is a flowchart showing this method. Encoded video data is input for each frame (step 201), and the encoding is performed using the correlation between frames for a frame encoded using the correlation between frames and without using motion compensation. The counted macroblocks are counted within a certain counting interval (steps 201 to 204), and the telop portion is selected based on the comparison result with the threshold value (step 205).
[0034]
FIG. 3 is a block diagram of the telop candidate image creation unit 102 based on this method. The frame discriminating unit 301 extracts a frame encoded using the correlation between the frames from the encoded video data. The counting unit 302 calculates, for each macroblock position, the number of macroblocks encoded using the correlation between frames and without using motion compensation, for each macroblock position. Count within. The counting results are input as a matrix to the selection unit 303, and the selection unit 303 outputs a matrix representing the telop area based on the comparison result with the given threshold value. Finally, the connected area creation unit 304 obtains a connected area of the telop area matrix, and outputs each connected area as a telop candidate image.
[0035]
FIG. 4 is a block diagram of a video telop detection apparatus according to the second embodiment of the present invention.
[0036]
In this embodiment, a mask creation unit 107 is added to the first embodiment. The mask creation unit 107 creates a mask using the telop candidate image and the difference image. The mask is an image in which the area for evaluating the difference image is set to “1” and the other areas are set to “0”. The area for evaluating the difference image may be “0”, the other area may be “1”, or may be a value other than “0” or “1”.
[0037]
The advantage of the present embodiment is that the region for evaluating the difference image can be further limited as compared to the first embodiment. For example, if the telop candidate image covers a wider area than the area where the telop actually exists, the mask creation unit 107 can limit the determination area to only the area where the telop actually exists. As a result, the influence of the background of the telop is reduced and more accurate telop detection is possible.
[0038]
FIG. 5 is a block diagram illustrating an example of the mask creation unit 107.
[0039]
The new candidate pixel detection unit 402 compares the telop candidate image input to the telop candidate image input terminal 401 with the telop candidate image in the previous frame, and extracts a newly appearing telop candidate pixel. Note that the frame interval for obtaining the change in the telop candidate image may be between adjacent frames or several frames apart, but the frame interval for obtaining the difference image is appropriate. On the other hand, the invariant pixel detection unit 404 extracts, from the difference image input to the difference image input terminal 403, only the pixel value that is smaller than a certain threshold value and the value of the corresponding telop candidate image is “1”. . As for both outputs of the new candidate pixel detection unit 402 and the invariant pixel detection unit 404, the OR circuit 405 synthesizes a mask in which at least one pixel is “1” and the other is “0”, and the mask output Output to terminal 406.
[0040]
FIG. 6 is a diagram showing a state of processing by the mask creation unit 107. 6 (1) and 6 (2) represent the telop candidate image and the pixels constituting the actual telop, and are temporally changed from FIG. 6 (1) to FIG. 6 (2). In the figure, the hatched portion surrounded by a solid line frame represents a portion that is a telop candidate, and the black region represents a pixel constituting an actual telop. Since the new candidate pixel detection unit 402 detects peripheral telop candidate pixels including the right “C” between FIG. 6A and FIG. 6B, the right side of the mask in FIG. In the section, the rectangular area is 1. Further, in FIG. 6A and FIG. 6B, if the change of the pixel in the background portion “B” is severe, only the pixels constituting the actual telop are detected by the invariant pixel detection unit 404. In the central portion of the mask in FIG. 6 (3), the portion corresponding to the pixels constituting the telop is “1”. In the portion “A”, since the telop has disappeared at the time of FIG. 6B, the corresponding mask is “0”. In this way, in a portion where the telop does not change, a portion corresponding to a pixel constituting the telop is set to “1”, and in a portion where a new telop appears, a portion where the telop candidate image is “1” is set to “1”. A mask of 1 ″ can be synthesized.
[0041]
FIG. 7 is a block diagram illustrating another example of the mask creation unit 107.
[0042]
The telop candidate image input terminal 501, the buffer 502, and the AND circuit 503 constitute a new candidate pixel detection unit 402. The telop candidate image input to the telop candidate image input terminal 501 is stored in the buffer 502 for one frame. The AND circuit 503 compares the telop candidate images between the current frame and the previous frame stored in the buffer 502. Here, a value of “1” is output when a telop candidate exists only in the current frame and does not exist in the previous frame, and a value of “0” is output otherwise.
[0043]
The difference image input terminal 504, the threshold value input terminal 505, the comparator 506, the AND circuit 507, and the buffer 508 constitute an invariant pixel detection unit 404. In the comparator 506, the difference image input from the difference image input terminal 504 is compared with the threshold value D input to the threshold value input terminal 505. The difference images are compared for each pixel, and a value of “1” is output when the threshold value D is larger, and a value of “0” is output otherwise. Further, the AND circuit 507 sets “0” for pixels that are not telop candidates. The result is stored in the buffer 508 for one frame.
[0044]
The output of the AND circuit 503 and the output of the buffer 508 are integrated by the OR circuit 509, and at least one of them is “1”. When both are “0”, the mask output is “0”. Is output.
[0054]
FIG. 11 is a flowchart showing an embodiment of the telop detection method of the present invention.
[0055]
First, in the image input stage 801, an image is input. Next, in the telop candidate image creation stage 802, a telop candidate image is detected, and a telop candidate image is created in which the pixel in which the telop candidate image exists is set to “1” and the other pixels are set to “0”. Also, in the difference image creation stage 803, a difference image is created with the absolute value of the difference between two frame images that are separated in time as pixel values. The frame interval may be between adjacent frames or may be an interval separated by a plurality of frames. Also, either the telop candidate image creation stage 802 or the difference image creation stage 803 may be executed first. In the extraction step 804, the portion of the difference image where the telop candidate image is “0” is set to “0”. In the determination step 805, the presence of a telop is determined.
[0056]
Before explaining the subsequent flowcharts, an image and a buffer to be processed in this embodiment will be explained.
[0057]
The input image is in an array called scr, and each pixel is represented by scr [p] indicated by the pixel number p. Further, it is assumed that a set of one or more pixels is a block and designated by a block number b. The telop candidate images are managed in units of blocks and are represented by tel [b]. If it is a telop candidate, tel [b] = 1, otherwise tel [b] = 0. Further, the telop candidate image of the frame in the previous processing stage is represented by pretel [b]. The difference image is managed in units of pixels and is represented by diff [p]. Similarly, the difference image of the frame in the previous processing stage is represented by prediff [b].
[0058]
FIG. 12 is a flowchart showing an example of the telop candidate image creation stage 802.
[0059]
First, in step 901, the block number b is initialized to 0. Next, in step 902, it is checked whether or not the block indicated by the block number b is a telop candidate. If it is a telop candidate, 1 is assigned to tel [b] in step 903, and if not, step 904 is performed. Then, 0 is substituted into tel [b]. For detecting a telop candidate image, a conventional technique, for example, a method based on the above-mentioned 1996 IEICE Information and System Society “Telop region detection method from MPEG video” can be used. Subsequently, in step 905, the value of b is incremented by 1. In step 906, it is determined whether the block number b is equal to the number of blocks, and the processing from step 902 is repeated until the block number b reaches the number of blocks. The telop candidate image tel [b] is created by the above procedure.
[0060]
FIG. 13 is a flowchart showing an example of the difference image creation stage 803.
[0061]
First, in step 911, the pixel number p is initialized to 0. Next, in step 912, the absolute value of the difference between the pixel value scr [p] of the current frame and the pixel value prescr [p] of the previous frame is stored in the pixel value diff [p] of the difference image. Subsequently, the value of p is incremented by 1 in step 913, and the processing from step 912 is repeated until the pixel number p becomes equal to the total number of pixels in step 914. The difference image diff [p] is created by the above procedure.
[0062]
14 and 15 are flowcharts showing an example of the extraction stage 804.
[0063]
First, the outline will be described. In this example, a set of pixels that are invariant pixels in a telop area image or newly become telop candidates is set as a determination area, and a difference image included in the determination area is extracted. Here, an invariant pixel represents a pixel having a small change in value between frames. Then, the area count of the determination region and the area cdiff of the large portion of the difference image in the determination region, that is, the portion where the frame image has changed are calculated and output. In the determination area, the invariant pixel is represented by a pixel m [p]. Note that m [p] is 0 at first.
[0064]
First, in step 921, count, cdiff, and pixel number p are initialized to zero. Hereinafter, processing is performed for all pixel numbers p. In step 922, a block number b corresponding to the pixel number p is obtained. In the condition judgment at step 923, the pixel is an invariant pixel (m [p] = 1), or the block of number b is a telop candidate (tel [b] = 1) and is not a telop candidate in the previous frame ( It is determined whether or not at least one of pretel [b] = 0) is satisfied, and if it is satisfied, the process proceeds to step 924, and otherwise, the process proceeds to step 927. In this condition determination, it is determined whether the pixel is at least one of an invariant pixel or a new telop candidate. In step 924, the count value is incremented by one. Further, in step 925, the pixel value of the difference image with the pixel number p is checked. If the pixel value is larger than the threshold value D, the process proceeds to step 926, and the value of cdiff is increased by 1. Otherwise, the process proceeds to step 927. In step 927, the value of the pixel number p is incremented by 1, and the processing from step 922 is repeated until the pixel number p reaches the total number of pixels in step 928. By the above procedure, the area count of the determination area where a telop candidate exists and the change of the frame image should be examined, and the area cdiff of the portion where the frame image has changed in the determination area can be obtained.
[0065]
Next, the invariant pixel m [p] is obtained by the processing of steps 929 to 935, and pretel and prescr are created. First, in step 929, the pixel number p is initialized to 0. In step 930, a block number b corresponding to p is obtained. In step 931, it is determined whether or not the block b is a telop candidate (tel [b] = 1) and the pixel value of the difference image number p is equal to or less than a threshold value D (diff [p] ≦ D). If is true, the process proceeds to step 932 and the value of m [p] is set to 1. Otherwise, the process proceeds to step 933 and the value of m [p] is set to 0. In step 934, the value of p is incremented by 1, and in step 935, the processing from step 930 is repeated until p reaches the total number of pixels. The invariant pixel m [p] can be obtained by the above procedure. Next, in step 936, the contents of tel and scr of the current frame are copied to pretel and prescr, respectively, to prepare for the next processing stage. Finally, the previously calculated count and cdiff are output in step 937, and this stage is completed.
[0066]
FIG. 16 is a flowchart illustrating an example of the determination stage 805. First, the outline will be described. In this example, an area count of a determination region to be examined for a change in the frame image and an area cdiff of a portion where the frame image has changed in the determination region are input. The variable t has an integer value and represents the duration of the telop. Count and cdiff are compared, and while the telop continues to appear, the value of t increases by one, and t is reset to 0 when there is no telop or when the telop is switched. In order to remove noise for a short time, it is recognized as a telop when it continues for a threshold value T or more. Therefore, T and t can be compared to determine “start”, “continuing”, and “end” of the telop, and further, sub-procedures corresponding to each can be executed. In this example, each sub-procedure is not particularly defined, and can be variously defined depending on the application. For example, a drawing function for displaying the detection state on the screen can be defined as a sub procedure.
[0067]
In step 941, it is checked whether t is larger than 0. When t is 0 or less, that is, when there is no telop at present, the process proceeds to step 942, and when t is larger than 0, that is, when telop exists, the process proceeds to step 944. In step 942, the value of count is compared with the threshold value A, and if it is equal to or larger than A, the process proceeds to step 943 and the value of t is set to 1. That is, it represents the start of a telop.
[0068]
In step 944, count is compared with threshold A, and the number of cdiff divided by count is compared with threshold R to determine whether at least one of count <A and cdiff/count> R is satisfied. Proceed to 946, otherwise proceed to step 945. In this condition determination, the case where there is no telop or the image change in the determination region is large is examined. In step 945, since the condition is not satisfied, the telop is continuing and the value of t is incremented by one.
[0069]
In step 946, it is indicated that the condition is satisfied and the telop has disappeared or switched, so it is checked whether the telop duration t is equal to or greater than the threshold T. If the condition is satisfied, the sub-procedure “telop start” to be described later is called, and the sub-procedure “telop end” in step 947 is executed so as to correspond to it. In either case, in step 948, the value of t is reset.
[0070]
The above is the procedure for changing the value of t based on the comparison result between count and cdiff. Subsequent steps 949 to 952 are processes for comparing t with the threshold value T and calling the sub-procedures “telop start” and “telop ongoing”. First, in step 949, it is checked whether t is equal to the threshold value T. If they are equal, the sub-procedure “start telop” in step 950 is executed. If they are not equal, it is checked in step 951 whether t is greater than T. If the condition is satisfied, the sub-procedure “telop ongoing” in step 952 is called.
[0071]
In addition, this invention is not limited to the above embodiment, In the range which does not change the meaning, it can implement variously. For example, a final telop presence determination can be used to output a frame image and create a telop list.
[0072]
【The invention's effect】
As described above, the present invention has the following effects.
(1) Claim 1 and 3 According to the invention, since the telop is detected based on the change of the frame image in the area where the telop exists, it is possible to detect the change of the telop.
(2) Claim 2 When 4 According to the invention, since the area for checking the change in the frame image can be limited to the area where the telop exists, it is less affected by the change in the telop background, and the telop switching can be detected more accurately. .
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a video telop detection device according to a first embodiment of the present invention.
FIG. 2 is a flowchart showing a telop candidate image detection method.
FIG. 3 is a block diagram showing a telop candidate image creation unit 102;
FIG. 4 is a block diagram showing a video telop detection device according to a second embodiment of the present invention.
5 is a block diagram illustrating an example of a mask creation unit 107. FIG.
FIG. 6 is an explanatory diagram for explaining processing of a mask creation unit 107;
FIG. 7 is a block diagram illustrating another example of a mask creation unit 107.
[Figure 8 It is a flowchart showing a video telop detection method according to an embodiment of the present invention.
[Figure 9 This is a flowchart showing a telop candidate image creation stage 802.
[Figure 10 This is a flow chart showing a difference image creation stage 803.
[Figure 11 A flow chart representing the extraction stage 804.
[Figure 12 A flow chart representing the extraction stage 804.
[Figure 13 This is a flow chart showing the determination stage 805.
[Explanation of symbols]
101 Video input terminal
102 Ticker candidate image creation section
103 Difference image creation unit
104 AND circuit
105 judgment part
106 Telop output terminal
107 Mask part
201-205 steps
301 Frame discriminator
302 Counting unit
303 Selector
304 Linked area creation unit
401 Ticker candidate image input terminal
402 New candidate pixel detection unit
403 Difference image input terminal
404 Invariant pixel detector
405 OR circuit
406 Mask output terminal
501 Telop candidate image input terminal
502 buffers
503 AND circuit
504 Difference image input terminal
505 Threshold input terminal
506 Comparator
507 AND circuit
508 buffer
509 OR circuit
510 Mask output terminal
801 Image input stage
802 Ticker candidate image creation stage
803 Difference image creation stage
804 Extraction stage
805 Judgment stage
901-906, 911-914, 921-937, 941-952 steps

Claims (7)

A method for detecting a telop from an image,
A telop candidate image creating stage for creating a telop candidate image indicating an area where a telop candidate exists from a frame image;
A difference image creation stage for creating a difference image of two frame images;
An extraction step of extracting only a portion where a telop candidate exists from the difference image;
A video telop detection method comprising a determination step of determining the presence of a telop based on the telop candidate image and the extracted difference image. A method for detecting a telop from an image,
A telop candidate image creating stage for creating a telop candidate image indicating an area where a telop candidate exists from a frame image;
A difference image creation stage for creating a difference image of two frame images;
Using the telop candidate image and the difference image, a mask creating step of creating a mask indicating a region in which a change in pixel value is to be detected;
Extracting only the portion of the mask from the difference image;
A video telop detection method comprising determining the presence of a telop based on the telop candidate image and the extracted difference image. An apparatus for detecting a telop from an image,
Telop candidate image creating means for creating a telop candidate image indicating an area where a telop candidate exists from a frame image;
Difference image creating means for creating a difference image of two frame images;
Extraction means for extracting only a portion where a telop candidate exists from the difference image;
A video telop detection apparatus comprising: determination means for determining the presence of a telop based on the telop candidate image and the extracted difference image. An apparatus for detecting a telop from an image,
Telop candidate image creating means for creating a telop candidate image indicating an area where a telop candidate exists from a frame image;
Difference image creating means for creating a difference image of two frame images;
Using the telop candidate image and the difference image, a mask creating means for creating a mask indicating a region where a change in pixel value should be detected;
Extracting means for extracting only a portion of the mask from the difference image;
A video telop detection apparatus comprising: determination means for determining the presence of a telop based on the telop candidate image and the extracted difference image. The telop candidate image creating means extracts frame encoded from the encoded video data using the correlation between frames, and correlates the frames among the macroblocks of the extracted frames. A counting means for counting the number of macroblocks coded without using motion compensation and for each macroblock position within a certain counting interval for each macroblock position is compared with a given threshold value. 5. The apparatus according to claim 3 , further comprising: selection means for outputting a matrix representing a telop area; and connected area creating means for obtaining a connected area of the matrix and outputting each connected area as a telop candidate image. The mask creation means includes an invariant pixel detection means for detecting an invariant pixel that is a pixel having a small value change between frames, a new candidate pixel detection means for detecting a newly appearing telop candidate, an invariant pixel and a new candidate pixel The video telop detection device according to claim 4, further comprising a mask composition unit configured to synthesize a mask indicating a region where at least one of the two exists. The new candidate pixel detection means compares the first buffer for storing one frame of the telop candidate image with the telop candidate image of the current frame and the telop candidate image stored in the first buffer. A first comparison unit that outputs a telop candidate image of the current frame when it exists only in the current frame, the invariant pixel detection unit; a second comparison unit that compares the difference image with a threshold; and the difference image When the threshold value is larger than the threshold, the second buffer for storing the current frame telop candidate image for one frame and the current frame telop candidate image output from the first comparison means or the second buffer are output as mask outputs. The apparatus of claim 6 including mask output means.

Images