JP2007074091A - Telop detection device and method therefor, and recorder - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a telop detection device and a telop detection method capable of detecting a telop in an image precisely, and to provide a recorder. <P>SOLUTION: The telop detector 113 detects the overlapped section between edge image data extracted from the video signal of the present frame in an input video signal, and those extracted from the video signal of at least the last frame as still edge image data; and determines a telop region according to the degree of congestion of an edge indicated by the still edge image data, when it is determined that the still edge image data continue with nearly the same content at least for a prescribed number of frames. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a telop detection device and method for detecting information of a telop portion from a video signal, and a recorder such as a disk recorder to which the telop detection device and method are applied.

  There are TV broadcast programs such as information programs and variety programs in which the entire program is composed of several topics. In such programs, telops are often displayed in the corners of the screen to indicate what topic the video being broadcast is. For example, a telop B is displayed in the upper right corner of the screen A as shown in FIG. Such a telop is called a corner telop (meaning “corner” = “screen corner”). The corner telop is for indicating the topic of the program, and can be said to be important key information for understanding the contents of the program. Therefore, if a corner telop can be detected, applications such as structuring the entire program for each topic and support for scene search by the user are possible.

  FIG. 2 shows the difference between such a corner telop and a normal telop added to the video being broadcast. A normal telop generally needs to make the telop stand out in order to attract attention from the viewer, the luminance difference between the pixels constituting the normal telop and the background is large, and the luminance intensity of edges (character edges) is also strong. In addition, the display time of a normal telop is short and is approximately several seconds.

  On the other hand, the corner telop is a telop indicating an outline of the topic in the program being broadcast as described above. Thereby, even when the user views the program from the middle, it is possible to roughly understand what the program is if the corner telop is viewed. After that, since it is basically an unnecessary telop, it is desirable that the display form does not interfere with the main video as much as possible. Therefore, the luminance difference between the pixels constituting the corner telop and the background is relatively small, and the edge strength is also weak. However, since the corner telop is displayed while the same topic continues, the display time of the corner telop is long and is displayed for at least several tens of seconds.

  As a conventional method for detecting a normal telop, Patent Document 1 discloses a method for detecting a breaking telop such as emergency news as a telop displayed at the end of the screen. According to this detection method, the breaking news telop is displayed in a predetermined area at the top or bottom of the TV screen. Therefore, a luminance change above the threshold on the time axis is detected for these screen areas, and the brightness above the threshold is detected. The time of change is detected as the start time and end time of the breaking news telop.

  Further, according to the telop detection method disclosed in Patent Document 2, since telops are generally high-luminance and have a strong edge at the boundary with the background, only high-luminance pixels are effective. The telop portion is detected by superimposing (AND processing) the luminance binarized image, which is the obtained image, and the edge expanded image obtained by expanding the edge detected image by several pixels.

Further, Patent Document 3 discloses a uniform recording area detection circuit for comparing a luminance level of an arbitrary pixel of a video signal in a television signal with a luminance level of peripheral pixels to detect an area having a substantially uniform luminance level, and an image. A high-frequency region detection circuit that detects a high frequency component in an arbitrary video plane of a signal, and a constant-time unchanged pixel detection circuit that determines whether or not the luminance level of the arbitrary pixel remains unchanged for a predetermined time or more; , And a method for extracting a telop based on each output of these circuits is disclosed.
JP-A-12-23062 Japanese Patent Laid-Open No. 12-182053 JP-A-10-233994

  However, as described above, corner telops are different from ordinary telops. Therefore, the disclosed techniques of Patent Documents 1 and 2 are effective in detecting telops having a large luminance difference from the background. As described above, there is a drawback that it is not suitable for detecting a telop having a small luminance difference from the background. The technique disclosed in Patent Document 3 is effective for detecting a telop in which the constituent luminance of characters is substantially uniform. However, corner telop has little guarantee that the constituent luminance is uniform, and the corner telop can be detected with high accuracy. Has the disadvantage of being difficult.

  The problems to be solved by the present invention include the above-mentioned drawbacks as an example, and to provide a telop detection device and method and a recorder capable of detecting a telop such as a corner telop in a video with high accuracy. It is an object of the invention.

  The telop detection device according to the first aspect of the present invention includes an edge image extraction unit that extracts edge image data indicating an edge on luminance for each frame from an input video signal, and the image signal of the current frame by the edge image extraction unit. Still edge detection means for detecting, as still edge image data, an overlapping portion between the extracted edge image data and at least the edge image data extracted from the video signal of the previous frame, and the still edge image data is a predetermined number of frames or more. And the continuation state determination means for determining whether or not the continuation state detection means continues and the continuation state detection means continues the still edge image data with substantially the same content over the predetermined number of frames. Is determined according to the edge density indicated by the determined still edge image data. It is characterized by comprising a determining telop determining means region.

  The recorder of the invention according to claim 9 is a recorder for recording an input video signal on a recording medium, and edge image data indicating an edge on luminance for each frame from the input video signal or the video signal recorded on the recording medium. An edge image extraction means for extracting the edge image data, and the edge image data extracted from the video signal of the current frame by the edge image extraction means and at least the edge image data extracted from the video signal of the previous frame A stationary edge detecting means for detecting as image data, a continuation state determining means for determining whether or not the stationary edge image data is continued with substantially the same content over a predetermined number of frames, and the continuation state detecting means. It is determined that the still edge image data continues with substantially the same content over the predetermined number of frames. When in is characterized by comprising a determining telop determining means telop area in accordance with the density of the edge indicated by the decision it is still edge image data.

  According to an eleventh aspect of the present invention, there is provided a telop detection device for extracting edge image data indicating an edge on luminance for each frame from an input video signal and dividing one frame into blocks each composed of a predetermined plurality of pixels. Edge block creation means for creating edge block image data representing a block including a predetermined number of edges according to the edge image data, and edge block image data of the frame created this time by the edge block creation means; Still block detecting means for detecting at least a portion overlapping with edge block image data of a previously created frame as the still block image data, and the still block image data continuing with substantially the same content over a predetermined number of frames Continuation state determining means for determining whether or not the continuation When it is determined by the state detection means that the still block image data has continued with substantially the same content over the predetermined number of frames, a telop is selected according to the block density indicated by the determined still block image data. And a telop determination means for determining an area.

  The recorder of the invention according to claim 17 is a recorder for recording an input video signal on a recording medium, and edge image data indicating an edge on luminance for each frame from the input video signal or the video signal recorded on the recording medium. An edge image extracting means for extracting the edge, and an edge for dividing one frame into a block unit composed of a predetermined plurality of pixels, and generating edge block image data representing a block including a predetermined number of edges according to the edge image data A block creating unit and a still block detecting unit for detecting, as the still block image data, an overlapping portion between the edge block image data of the frame created this time by the edge block creating unit and the edge block image data of at least the previously created frame And the still block image data exceeds a predetermined number of frames. And the continuation state determination means for determining whether or not the continuation state detection means continues and the continuation state detection means continues the still block image data with substantially the same content over the predetermined number of frames. And a telop determination means for determining a telop area in accordance with the block density indicated by the determined still block image data.

  In the telop detection method according to the nineteenth aspect of the present invention, edge image data indicating an edge in luminance is extracted for each frame from the input video signal, and the edge image data extracted from the video signal of the current frame and at least the previous frame are extracted. A portion overlapping with the edge image data extracted from the video signal is detected as the still edge image data, and it is determined whether or not the still edge image data continues with substantially the same content over a predetermined number of frames. When it is determined that the still edge image data continues with substantially the same content over the predetermined number of frames, the telop area is set according to the edge density indicated by the determined still edge image data. It is characterized by judging.

  A telop detection method according to a twentieth aspect of the present invention extracts edge image data indicating an edge on luminance for each frame from an input video signal, divides one frame into blocks each consisting of a predetermined plurality of pixels, and the edge image Create edge block image data that represents a block that contains a predetermined number of edges or more according to the data, and create an overlap between the edge block image data of the currently created frame and at least the edge block image data of the previously created frame. It is detected as the still block image data, and it is determined whether or not the still block image data continues with substantially the same content over a predetermined number of frames, and the still block image data exceeds the predetermined number of frames. If it is determined that the content has been continued with substantially the same content, the determined still block image data is determined. It is characterized by determining the telop area in accordance with the density of the block indicated by the data.

  In the telop detection device according to the first aspect of the present invention, there is an overlapping portion between the edge image data extracted from the video signal of the current frame of the input video signal and the edge image data extracted from at least the video signal of the previous frame. If it is detected as still edge image data and the still edge image data continues with almost the same content over a predetermined number of frames, the telop area is determined according to the edge density indicated by the still edge image data. The

  In the recorder of the invention according to claim 9, the edge image data extracted from the video signal of the current frame of the input video signal or the video signal recorded on the recording medium and at least extracted from the video signal of the previous frame If the overlapping portion with the edge image data is detected as still edge image data and the still edge image data continues with substantially the same content over a predetermined number of frames, the density of edges indicated by the still edge image data The telop area is determined according to the above.

  In the telop detection method according to the nineteenth aspect of the present invention, the overlap between the edge image data extracted from the video signal of the current frame of the input video signal and the edge image data extracted from at least the video signal of the previous frame. When a portion is detected as still edge image data, and it is determined that the still edge image data continues with almost the same content over a predetermined number of frames, a telop is selected according to the edge density indicated by the still edge image data. An area is determined.

  Therefore, in the inventions according to claims 1, 9, and 19, it is possible to detect a telop having a relatively small luminance difference between the pixels constituting the telop and the background and having a weak edge luminance intensity with high accuracy. it can.

  In the telop detection device according to an eleventh aspect of the present invention, edge block image data representing a block in which one frame is divided into blocks each including a plurality of predetermined pixels and the number of edges is greater than or equal to a predetermined number according to the edge image data. Created for each frame of the input video signal, the overlapping part of the edge block image data of the frame created this time and at least the edge block image data of the previously created frame is detected as still block image data, and the still block image data is If the same content is continued for a predetermined number of frames or more, the telop area is determined according to the block density indicated by the still block image data.

  In the recorder according to the seventeenth aspect of the present invention, one frame of the input video signal or the video signal recorded on the recording medium is divided into blocks each composed of a predetermined plurality of pixels, and the number of edges is determined according to edge image data. Edge block image data representing a block containing a predetermined number or more is created for each frame of the input video signal, and the overlapping part between the edge block image data of the currently created frame and at least the edge block image data of the previously created frame is stationary. If the block image data is detected and the still block image data continues with substantially the same content over a predetermined number of frames, the telop area is determined according to the block density indicated by the still block image data. .

  In the telop detection method according to the twentieth aspect of the present invention, edge block image data representing a block including a predetermined number of edges or more according to edge image data is obtained by dividing one frame into blocks each composed of a predetermined plurality of pixels. And the overlapping portion of the edge block image data of the currently created frame and at least the edge block image data of the previously created frame is detected as still block image data, and the still block image data exceeds a predetermined number of frames. When it is determined that the continuation is substantially the same, the telop area is determined according to the block density indicated by the still block image data.

  Therefore, in the inventions according to claims 11, 17, and 20, it is possible to detect a telop having a relatively small luminance difference between the pixels constituting the telop and the background and having a weak edge luminance intensity with high accuracy. it can. In addition, by performing detection processing in units of blocks, telops can be detected with a small amount of data, so that the processing load can be reduced.

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

  FIG. 3 shows a disk recorder to which the telop detection device of the present invention is applied. This disk recorder includes a TV tuner 101, an A / D converter 102, an MPEG encoder 103, a control unit 104, a display unit 105, an operation unit 106, an HDD (hard disk drive) device 107, an optical disk drive device 108, a ROM 109, a RAM 110, and an MPEG. It comprises a decoder 111, a D / A converter 112, and a corner telop detection device 113.

  The TV tuner 101 is connected to an antenna input terminal 121, selectively receives a television broadcast wave through an antenna (not shown) connected thereto, under the control of the control unit 104, demodulates the received signal, and receives an analog signal. Output video and audio signals. The output of the TV tuner 101 is connected to the A / D converter 102. The A / D converter 102 digitizes the analog video signal and audio signal output from the TV tuner 101. The A / D converter 102 is connected to external video and audio input terminals 122 and 123, and digitizes analog video and audio signals supplied thereto. The output of the A / D converter 102 is connected to the encoder 103. The encoder 103 converts the digitized video signal and audio signal into, for example, MPEG format compressed data.

  The encoder 103, the control unit 104, the HDD device 107, the optical disk drive device 108, the ROM 109, the RAM 110, the decoder 111 and the corner telop detection device 113 are connected to a common bus 116.

  The control unit 104 is composed of, for example, a microcomputer, and controls the entire disk recorder according to a predetermined control program. The control unit 104 and the TV tuner 101 are connected by a control line (not shown), and the reception channel of the TV tuner 101 is controlled by the control unit 104.

  The display unit 105 and the operation unit 106 are connected to the control unit 104. The display unit 105 displays the reception channel of the TV tuner 101 and the recording / reproduction status of the HDD device 107 and the optical disc drive device 108 in response to a display command from the control unit 104. The operation unit 106 accepts a user input operation and supplies a command corresponding to the input operation to the control unit 104.

  The HDD device 107 is a storage device using a rewritable fixed disk (not shown), and the write instruction of the internal fixed disk indicates the compressed data supplied from the encoder 103 in accordance with the write instruction of the control unit 104. Write to address location. Further, in accordance with a read instruction from the control unit 104, the compressed data at the storage location addressed by the read instruction is read from the fixed disk and output to the bus 116.

  The optical disk drive device 108 is a removable storage device and can write compressed data from the encoder 103 or the HDD device 107 to the optical disk 115 in accordance with a write instruction from the control unit 104. For example, DVD-R or DVD-RW is used as the optical disc 115. The optical disk drive device 108 can also read the compressed data written on the optical disk 115 in response to a read instruction from the control unit 104.

  The control program is written in the ROM 109 in advance. Data required for the control operation of the control unit 104 is written into the RAM 110 or data is read therefrom.

  The decoder 111 decompresses the compressed data supplied from the encoder 103, the control unit 104, the HDD device 107, or the optical disk drive device 108 via the bus 116, and restores a digital video signal and audio signal. The restored digital video signal and audio signal are supplied to the D / A converter 112. The D / A converter 112 converts digital video signals and audio signals into analog video signals and audio signals, and supplies them to video and audio output terminals 124 and 125. A television monitor (not shown) is connected to the video and audio output terminals 124 and 125.

  The corner telop detection device 113 is connected to the bus 116 as described above, and detects a corner telop according to the compressed data in the MPEG format. In the corner telop detection device 113, either the compressed data output from the encoder 3, the compressed data read from the HDD device 107, or the compressed data read from the optical disc 115 by the optical disc drive device 108 is controlled. Supplied under the control of 104. The corner telop detection output from the corner telop detection device 113 is supplied to the control unit 104.

  FIG. 4 shows a schematic configuration of the corner telop detection device 113. The corner telop detection device 113 includes an edge extraction unit 1, a corner telop candidate detection unit 2, and a corner telop determination unit 3.

  The edge extraction unit 1 includes a frame extraction unit 11, an in-frame edge extraction unit 12, a stationary edge detection unit 13, and a stationary edge buffer 14, and detects and stores a stationary edge for each frame from the input video signal. The frame extraction unit 11 extracts a video signal for one frame from the input video signal. The in-frame edge extraction unit 12 extracts edge image data composed of pixels in which an edge on luminance is formed from a video signal for one frame. The edge image data indicates a sharp portion where the luminance difference with an adjacent pixel is a predetermined value or more, that is, an edge such as an edge of a telop character. The stationary edge detection unit 13 compares the extracted edge image data with the already detected stationary edge image data, and detects the edge pixel of the portion whose position does not change as the stationary edge image data. The stationary edge buffer 14 stores still edge image data detected by the stationary edge detector 13.

  The corner telop candidate detection unit 2 includes an edge number measurement unit 21 and a duration measurement unit 22, and detects how many frames the still edge image data detected by the still edge detection unit 13 has continued. The edge number measurement unit 21 measures the number of edges indicated by the still edge image data detected by the still edge detection unit 13. The duration measurement unit 22 measures the number of frames in which the state in which the number of edges measured by the edge number measurement unit 21 is equal to or greater than the threshold value Tn continues.

  The corner telop determination unit 3 includes an edge block creation unit 31, a block density calculation unit 32, and an edge block buffer 33, and determines a corner block region based on still edge image data. The edge block creation unit 31 creates still edge block image data according to still edge image data. The block density calculation unit 32 calculates the block density in the image and determines the corner block region according to the density. The edge block buffer 33 holds edge block image data.

  In this embodiment, unless otherwise specified, the elements constituting the pixels of each frame are luminance signals.

  The operation of the corner telop detection device 113 having such a configuration will be described below with reference to flowcharts.

  As shown in FIG. 5, in the edge detection unit 1, before accepting an input video signal, a frame count value count indicating the number of continued frames of a corner telop candidate is reset to an initial value 0 as an initialization process (Step S1). S1). When the actual corner telop detection process is started, first, the frame extraction unit 11 extracts a video signal (pixel data) for one frame from the input video signal (step S2). The input video signal is a digital compression signal in a compression format such as MPEG, but other analog signals, digital signals, and digital compression signals in other compression formats may be used. The image size is not limited, but an analog video signal is supplied to the frame extraction unit 11 via an A / D (analog / digital) converter.

  The frame extraction unit 11 divides one frame into a plurality of areas and divides the frame into an effective part and an invalid part (step S3). For example, as shown in FIG. 6, one extracted frame is divided into a total of 8 areas of 4 in the vertical direction and 2 in the horizontal direction, and 4 areas corresponding to the corners of the screen are set as effective areas, and the rest are set as invalid areas. Since the corner telop has a high tendency to be displayed at the corner of the screen, it is expected that the processing efficiency and detection accuracy can be improved by limiting the effective area in this way. In the case of FIG. 6, the subsequent edge processing and the like are performed individually for the four effective areas. Also, the region dividing method may be other than this, and the region dividing itself may not be performed.

  The in-frame edge extraction unit 12 extracts edges from the pixel data of the effective area in the frame, and creates edge image data (step S4). As a filter used for edge extraction, for example, a Laplacian filter can be used, but is not particularly limited thereto. Since the corner telop has the property that the edge is weak, the edge determination may be preferably relaxed.

  The still edge detection unit 13 determines whether the current frame is the first frame of the input video signal or a frame immediately after the corner telop candidate is detected (step S5). That is, it is determined whether or not the frame count value count is zero. If the frame in which the edge is detected in step S4 is the first frame or the frame immediately after the corner telop candidate is detected, count = 0, and the detected edge image data is set as the still edge image data and is stored in the still edge buffer 14. Store (step S6). On the other hand, if the frame in which the edge detection is performed in step S4 is not the first frame or the frame immediately after the corner telop candidate detection, the edge image data created this time and the still edge image data already detected (for example, the previous detection) AND (logical product) with the still image edge image data) is taken as new still edge image data and stored in the still edge buffer 14 (step S7). In step S7, not only the AND process of the edge image data created this time and the already detected still edge image data, but the OR (logical sum) of the edge image data created this time and the already detected still edge image data is performed. ) And may be used as still edge image data, or other superposition methods may be used.

  For example, when a corner telop image as shown in FIG. 7A is in the effective area, the still edge image data is as shown in FIG. 7B.

  The edge number measuring unit 21 acquires still edge image data from the still edge buffer 14, measures the number of edges indicated by the still edge image data, and determines whether or not the number of edges is equal to or greater than a threshold value Tn (first threshold value). Is discriminated (step S8). When the number of edges is equal to or greater than the threshold value Tn, the frame count value count is incremented by 1 (step S9), and it is determined whether or not the processing of all the frames of the input video signal has been completed (step S10). If all the frames have not been processed, the processing of the input video signal is advanced to the next frame (step S11), and the process returns to step S2 to execute the corner telop detection processing for the next frame. .

  When the number of edges becomes equal to or greater than the threshold value Tn, the duration measurement unit 22 determines whether or not the frame count value count is greater than the threshold value Tc (second threshold value) (step S12). The threshold value Tc corresponds to the lower limit of the corner telop display time. Since the corner telop has the property that the display time is long, the threshold value Tc may be preferably set longer. As a specific method for setting the threshold value Tc, it is conceivable to set the time of each topic in the program by analogy based on the overall length of the program. For example, when the program length is 1 hour, assuming that the number of topics is about 10, the average time of each topic is about 6 minutes. However, since the program actually includes commercials in addition to the topic, Tc may be set to about a few minutes, for example. Further, based on the nature of each program, Tc may be set longer if the number of topics is small, and conversely, Tc may be set shorter if the number of topics is large. For all programs, the threshold Tc may be set to a fixed length such as 1 minute.

  If count ≦ Tc, there is a high possibility that the stored still edge image data is not a corner telop image. Therefore, the still edge image data stored in the still edge buffer 14 is deleted, and the frame count value count is the initial value. It is reset to 0 (step S13). Thereafter, step S10 is executed.

  If count> Tc, the corner telop determination subroutine by the corner telop determination unit 3 is executed (step S14), and then step S13 is executed.

  In the corner telop determination subroutine, as shown in FIG. 8, first, the edge block creation unit 31 creates edge block image data from the still edge image data detected by the corner telop candidate detection unit 2, and the edge block image data Is stored in the edge block buffer 33 (step S21). The edge block image data indicates an image obtained by dividing a still edge image into blocks each composed of 8 × 8 pixels and using a block having a large number of edges in which the number of edges in each block is a predetermined number or more as an edge block. For example, in the still edge image data image shown in FIG. 7B, the block unit is indicated by a broken line as shown in FIG. 7C, and the edge block is black as shown in FIG. 7D. It is a block. Note that the block is not limited to 8 × 8 pixels, and the vertical and horizontal directions may be composed of other numbers of pixels.

  The edge block creation unit 31 removes spatially isolated blocks from the edge block image (step S22). As shown in FIG. 9 (a), when a still edge image includes an edge such as dust other than the corner telop, the edge block image based on the edge includes a corner telop portion as shown in FIG. 9 (b). Other blocks will appear. Therefore, in step S22, as shown in FIG. 9C, an edge block image from which such an isolated block is removed is created.

  The block density calculation unit 32 determines whether or not the number of blocks of the edge block image created by the edge block creation unit 31 is larger than the threshold Tb (third threshold) (step S23). If the number of blocks is equal to or less than the threshold value Tb, the corner telop determination subroutine is immediately terminated.

  If the number of blocks is larger than the threshold value Tb, the block density calculation unit 32 sets the minimum rectangular area surrounding the edge block as the block surrounding area (step S24). In the edge block image created by the edge block creation unit 31, for example, as shown in FIG. 10 (a), if an edge block exists, a rectangular block surrounding area as shown by a bold line in FIG. Is set. The block density calculation unit 32 calculates the area ratio of the edge block with respect to the set block surrounding area, and determines whether or not the area ratio is equal to or greater than a threshold Ta (fourth threshold) (step S25). If the area ratio is smaller than the threshold Ta, the corner telop determination subroutine is immediately terminated.

  On the other hand, if the area ratio is equal to or greater than the threshold value Ta, the block density calculation unit 32 determines that the set block surrounding area includes a corner telop and outputs corner telop information (step S26). The corner telop information is at least one of the corner telop start time and end time, the still edge image, the edge block image, and the coordinates of the block surrounding area. Character recognition of a still edge image may be performed by OCR (optical character reader) processing and the result may be output as a character code.

  FIG. 11 shows an embodiment of the present invention. The corner telop detection device shown in FIG. 11 includes an edge extraction unit 1, a corner telop candidate detection unit 2, and a corner telop determination unit 3 as in the device of FIG. The edge extraction unit 1 and the corner telop candidate detection unit 2 are the same as those in the apparatus of FIG. The corner telop determination unit 3 includes an isolated edge removal unit 35 and an edge density calculation unit 36. The isolated edge removing unit 35 removes isolated edges from the still edge image. The edge density calculation unit 36 calculates the edge density in the still edge image from which the isolated edges are removed.

  In the corner telop determination subroutine by the corner telop determination unit 3 in FIG. 11, as shown in FIG. 12, the isolated edge removal unit 35 detects an isolated edge spatially isolated from the still edge image detected by the corner telop candidate detection unit 2. It is removed (step S31). For example, with respect to the still edge image including the isolated edge shown in FIG. 13A, the isolated edge removing unit 35 creates a still edge image from which the isolated edge is removed as shown in FIG. 13B. .

  The edge density calculation unit 36 determines whether or not the number of edges of the still edge image after the isolated edge removal is larger than the threshold value Te (fifth threshold value) (step S32). If the number of edges is less than or equal to the threshold Te, the corner telop determination subroutine is immediately terminated.

  If the number of edges is larger than the threshold value Te, the smallest rectangular area that surrounds all the edges of the still edge image is set as the edge surrounding area (step S33). In the still edge image after the isolated edge removal in FIG. 13B described above, for example, a rectangular edge surrounding area is set as indicated by a thick line in FIG. 13C.

  The edge density calculation unit 36 calculates the area ratio of the edge portion with respect to the set edge surrounding region, and determines whether or not the area ratio is equal to or greater than a threshold Ta (sixth threshold) (step S34). If the area ratio is smaller than the threshold Ta, the corner telop determination subroutine is immediately terminated.

  On the other hand, if the area ratio is equal to or greater than the threshold value Ta, the edge density calculation unit 36 determines that the set edge surrounding area includes a corner telop, and outputs corner telop information (step S35).

  In the embodiment shown in FIGS. 11 and 12, since the corner telop area can be determined on a pixel basis, the corner telop area can be determined more limitedly.

  FIG. 14 shows still another embodiment of the present invention. The corner telop detection device shown in FIG. 14 includes a stationary block detection unit 5, a corner telop candidate detection unit 6, and a corner telop determination unit 7. The still block detection unit 5 includes a frame extraction unit 51, an in-frame edge extraction unit 52, an edge block creation unit 53, a still block detection unit 54, and a still block buffer 55, and generates a still block image for each frame from an input video signal. Detect and save it.

  The frame extraction unit 51 extracts a video signal for one frame from the input video signal. The in-frame edge extraction unit 52 extracts an edge image composed of pixels that form edges from the video signal for one frame. The edge block creation unit 53 creates an edge block image from the extracted edge image. The still block detector 54 detects a still block image from the edge block image and stores it in the still block buffer 55.

  The corner telop candidate detection unit 6 includes a block number measurement unit 61 and a duration measurement unit 62, and detects how many frames the state of the still block image detected by the still block detection unit 13 has continued. The block number measuring unit 61 measures the number of blocks of the still block image detected by the still block detecting unit 13. The duration measuring unit 62 measures the number of frames in which the state in which the number of blocks measured by the block number measuring unit 61 is equal to or greater than the threshold value continues.

  The corner telop determination unit 7 includes an isolated block removal unit 71 and a block density calculation unit 72, and determines a block including a corner telop of a still block image. The isolated block removing unit 71 removes isolated blocks from the still block image. The block density calculation unit 72 calculates the block density in the still block image from which the isolated block is removed, and determines a block including a corner telop based on the block density.

  The operation of the corner telop detection device of FIG. 14 will be described below using a flowchart.

  As shown in FIG. 15, in the still block detection unit 5, before the input video signal is received, the frame count value count indicating the number of continuing frames of the corner telop candidate is reset to the initial value 0 as an initialization process ( Step S51). When the actual corner telop detection process is started, first, the frame extraction unit 51 extracts a video signal (pixel data) for one frame from the input video signal (step S52).

  The frame extraction unit 51 divides one frame into a plurality of areas and divides the frame into an effective part and an invalid part (step S53). After execution of step S53, the in-frame edge extraction unit 52 extracts an edge from the pixel data of the effective area and creates an edge image (step S54). Steps S51 to S54 are the same as steps S1 to S4 described above.

  The edge block creation unit 53 creates an edge block image from the edge image (step S55). As an edge block image creation method, the method shown in the edge block creation unit 31 is used. In other words, an edge image is divided into blocks each composed of 8 × 8 pixels, and an image is created in which blocks having a large number of edges in which the number of edges in each block is a predetermined number or more are used as edge blocks. Here, when the input video signal is a compressed video signal such as MPEG, a high frequency block (a block with many edges) can be detected based on the DCT coefficient. Such a block may be used as an edge block. In the case of a compressed video signal, the full decoding process of the input compressed video signal and the creation of the edge image in step S54 are not required, leading to a reduction in processing load. As a method for detecting a high-frequency block from MPEG DCT coefficients, for example, the method shown in the 59P0262 publication (please inform us of the publication number) can be used.

  The still block detection unit 54 determines whether or not the current frame in which the edge block image is generated by the edge block generation unit 53 is the first frame of the input video signal or a frame immediately after the corner telop candidate is detected (step). S56). If the current frame is a head frame or a frame immediately after corner telop candidate detection, the created edge block image is set as a still block image and stored in the still block buffer 55 (step S57). On the other hand, if the current frame is not the first frame or the frame immediately after the detection of the corner telop candidate, the AND (logic) of the edge block image created this time and the already detected still block image (for example, the previously detected still block image) Product) to make a still block image, which is stored in the still block buffer 55 (step S58). In step S58, an OR (logical sum) between the edge block image created this time and the already detected still block image is taken and used as a still block image, or a superposition method other than this may be used. .

  The block number measuring unit 61 reads out and acquires a still block image from the still block buffer 55, measures the number of blocks in the still block image, and determines whether or not the number of blocks is equal to or greater than a threshold value Tn (first threshold value). A determination is made (step S59). When the number of blocks is equal to or greater than the threshold value Tn, the frame count value count is increased by 1 (step S60), and it is determined whether or not the processing of all the frames of the input video signal has been completed (step S61). If all the frames have not been processed, the processing of the input video signal is advanced to the next frame (step S62), and the process returns to step S52 to execute the corner telop detection processing for the next frame. .

  When the number of blocks is equal to or greater than the threshold value Tn, the duration measurement unit 62 determines whether or not the frame count value count is greater than the threshold value Tc (second threshold value) (step S63). If count ≦ Tc, there is a high possibility that the still block image data currently stored in the still block buffer 55 is not corner telop image data. Therefore, the still block image data stored in the still block buffer 55 is erased, and The frame count value count is reset to the initial value 0 (step S64). Thereafter, step S61 is executed.

  If count> Tc, the corner telop determination subroutine by the corner telop determination unit 3 is executed (step S65), and then step S64 is executed.

  In the corner telop determination subroutine of step S65, as shown in FIG. 16, first, the isolated block removing unit 71 removes spatially isolated blocks from the still block image (step S71). In step S71, a still block image from which isolated blocks have been removed is created.

  The block density calculation unit 72 determines whether or not the number of blocks of the still block image after the isolated block removal by the isolated block removal unit 71 is larger than the threshold Tb (third threshold) (step S72). If the number of blocks is equal to or less than the threshold value Tb, the corner telop determination subroutine is immediately terminated.

  If the block density is greater than the threshold value Tb, the block density calculation unit 72 sets the minimum rectangular area surrounding the block of the still block image as the block surrounding area (step S73). The block density calculation unit 72 calculates the area ratio of the stationary block with respect to the set block surrounding area, and determines whether or not the area ratio is equal to or greater than a threshold Ta (fourth threshold) (step S74). If the area ratio is smaller than the threshold Ta, the corner telop determination subroutine is immediately terminated.

  On the other hand, if the area ratio is equal to or greater than the threshold value Ta, the block density calculation unit 72 determines that the set block surrounding area includes a corner telop and outputs corner telop information (step S75).

  In the embodiment shown in FIGS. 14 to 16, still block image data in units of blocks is detected instead of still edge image data in units of pixels, and corner telop candidate detection and corner telops are detected according to the still block image data. A determination is made. Therefore, since the storage capacity of the stationary block buffer 55 is smaller than the storage capacity of the stationary edge buffer 14 in FIG. 2, it is possible to easily implement even a device with few resources such as a built-in memory. In addition, when a compression encoded signal such as MPEG is used as the input video signal, an edge block (high frequency block) can be detected based on the DCT coefficient, so that the in-frame edge extraction unit 52 and the edge block creation unit in FIG. 53 becomes unnecessary, and the processing load can be reduced.

  In each of the above-described embodiments, a corner telop displayed mainly in an information program is detected, but it can also be applied to score display detection in a sports program such as baseball or soccer. The score display of a sports program is also displayed in the corner of the screen and has a property similar to that of a corner telop that is displayed during play. Therefore, it can be basically detected with the same block configuration as in FIG. However, depending on the type of sport, the following measures may be taken. Baseball and soccer are examples of typical sports programs. In addition to the score, the baseball score display displays a lot of information such as strike / ball / out count, and has strong edge strength. Also, the display time for one time is relatively short (several seconds). On the other hand, in the case of soccer, the edge strength is weak and the display time tends to be relatively long. The reason why the display time of the score differs depending on the sport is that each play unit (time length between play breaks) is different. In baseball, one pitching by the pitcher is basically one play unit. In soccer, one play unit is from the kick-off (or resumption of play) until the play is interrupted due to a foul or goal. Therefore, when the input video is a baseball video, the edge determination threshold Tn in the in-frame edge extraction unit 12 in FIG. 2 is set high, and the duration threshold Tc of the duration measurement unit 22 is set short, thereby It is possible to realize processing suitable for the score detection. On the other hand, in the case of soccer or the like, it is suitable for soccer score detection by setting the edge determination threshold Tn in the in-frame edge extraction unit 12 to be lower and setting the duration threshold Tc in the duration measurement unit 22 to be longer. Processing can be realized. Further, in the detection process of the score display, by comparing the similarity with the score data detected most recently in time, for example, it is possible to detect a place where a score has been entered. As a specific comparison method, a method of superimposing the current and previous still edge images and determining based on the degree of change is conceivable. This is effective for corner telop detection of not only sports programs but also information programs, and the change in the topic may be determined by comparing with the previously detected data.

  In the above-described embodiment, when one frame region is divided as shown in FIG. 6, there are a plurality of corner telop display candidates (four in this example). It is necessary to judge whether it is displayed. Thus, the corner telop display area may be determined based on the corner telop candidate detection frequency in each area. As one method, an area in which the most corner telop display candidates are detected among the four areas may be determined as the corner telop display area. However, in this case, for example, as shown in FIGS. 17A and 17B, there is a possibility that the time display at the corner of the screen or the telop of the broadcast station name is erroneously determined as a corner telop. In view of this, it is possible to add a process of excluding telops that change exactly every minute, or conversely, telops that do not change for a very long time, from corner telop candidates to prevent erroneous detection of corner telops. Further, if the program to be detected is a series program and a corner telop is detected from the previous broadcast, the display area of the corner telop may be determined based on the information.

  Further, in each of the above-described embodiments, the case where the present invention is applied to a corner telop detection device that detects a corner telop has been described, but the telop detected by the present invention is not limited to a corner telop. That is, the telop is often displayed as a corner telop at the corner of the screen, but is not necessarily displayed at the corner of the screen.

  As described above, since the edge detection unit 1, the corner telop candidate detection unit 3, and the corner telop determination unit 3 are provided, the luminance intensity of the edge is weak and the luminance difference between the pixels constituting the telop and the background is small. Corner telop can be detected with high accuracy. In addition, since the stationary block detection unit 5, the corner telop candidate detection unit 6, and the corner telop determination unit 7 are provided, not only can the corner telop be detected with high accuracy, but also detection processing is performed in units of blocks, so that there are few. Since the corner telop can be detected by the amount of data, the processing load can be reduced.

  The telop detection apparatus and method according to the present invention are for receiving a TV broadcast and recording or viewing, for example, a TV receiver such as a DVD / HDD recorder or a PDP TV apparatus, a PC (personal computer) with built-in TV tuner, or a car navigation system. The present invention can be applied to video equipment such as a device and software for a PC.

  A chapter menu can be created using the corner telop detected by the telop detection apparatus or method of the present invention. The same application is possible for all devices such as the above-mentioned DVD / HDD recorder, PDP television device, car navigation device and the like that can record and view video content. For example, in the case of the recorder shown in FIG. 3, first, the content is stored in the recorder's HDD device 107 (recorded compressed data), the content recorded on the optical disc 115, or the content to be recorded from now on. Detect telop. The corner telop detection method according to the present invention may be applied to either real-time detection in which content is simultaneously detected while recording or non-real-time detection in which recorded content is detected again. When corner telop is detected, at least the display start time of the telop is detected as corner telop information. If a corner telop is detected, the control unit 104 sets a chapter at the detected location to indicate a video break. In chapter setting, a chapter table file indicating the relationship between corner telop display start time and playback address is created. The chapter is set at a timing (for example, after recording) when the time information of the content and the video information are synchronized. Next, the control unit 104 creates a chapter menu using the detected corner telop as a button image. The chapter menu, as shown in FIG. 18, shows the contents of the chapter in a menu format in an easy-to-understand manner for the user. By viewing this chapter menu, the user can grasp the overall outline of the program at a glance. Furthermore, the display start time is specified by selecting a portion of interest from the menu, so that the playback address corresponding to that time is read from the chapter table file, and video playback from that playback address starts. Is done. Therefore, the user can view the video content from a favorite location. By applying the corner telop to content chaptering in this way, the content can be automatically structured according to the content. Furthermore, by creating a chapter menu using the corner telop as a button image, the user can grasp the entire outline of the program at a glance and can view the content from an interesting place. By applying the corner telop in this way, the convenience of content can be enhanced for the user.

It is a figure which shows the example of a display of a corner telop. It is a figure which shows the characteristic difference between a normal telop and a corner telop. It is a block diagram which shows the structure of the disc recorder to which this invention was applied. It is a block diagram which shows the structure of the corner telop detection apparatus in the recorder of FIG. It is a flowchart which shows operation | movement of the corner telop detection apparatus of FIG. It is a figure which shows several division area of 1 frame, its effective area | region, and an invalid area | region. It is a figure which shows a frame image, an edge image, the division | segmentation of a block unit, and an edge block image. It is a flowchart which shows a corner telop determination subroutine. It is a figure which shows the removal of an edge image, an edge block image, and an isolated block. It is a figure which shows an edge block image and a block surrounding area | region. It is a block diagram which shows the other Example of this invention. It is a flowchart which shows the corner telop determination subroutine performed with the corner telop detection apparatus of FIG. It is a figure which shows an edge image, the removal of an isolated block, and an edge surrounding area | region. It is a block diagram which shows the Example of this invention. It is a flowchart which shows operation | movement of the corner telop detection apparatus of FIG. It is a flowchart which shows the corner telop determination subroutine performed with the corner telop detection apparatus of FIG. It is a figure which shows the example of the time display in a video | video, and a broadcast station name telop. It is a figure which shows the example of a chapter menu.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Edge extraction part 2,6 Corner telop candidate detection part 3,7 Corner telop determination part 5 Still block detection part

Claims (20)

Edge image extraction means for extracting edge image data indicating an edge on luminance for each frame from an input video signal;
Still edge detection means for detecting, as still edge image data, an overlapping portion between the edge image data extracted from the video signal of the current frame by the edge image extraction means and the edge image data extracted from the video signal of the previous frame at least. When,
Continuation state determination means for determining whether or not the still edge image data has continued with substantially the same content over a predetermined number of frames;
When it is determined by the continuation state detection means that the still edge image data has continued with substantially the same content over the predetermined number of frames, according to the edge density indicated by the determined still edge image data And a telop determining means for determining a telop area.   The edge image extraction means includes a frame extraction unit that extracts a video signal for one frame of the input video signal, and a luminance difference between adjacent pixels from the video signal for one frame extracted by the frame extraction unit. The telop detection device according to claim 1, further comprising: an in-frame edge extraction unit that extracts edge video data indicating edge pixels having a luminance equal to or higher than a predetermined value. The edge image extracting means includes area dividing means for dividing one frame into a plurality of areas, making only the area of the screen corner an effective area, and making the other area an invalid area,
The intra-frame edge extraction unit is an edge video data indicating edge pixels having a luminance difference that is equal to or greater than a predetermined value from the luminance of adjacent pixels from the video signal of the effective area for one frame extracted by the frame extraction unit. The telop detection device according to claim 2, wherein the telop is extracted.   The still edge detection means includes a still edge buffer for storing the still edge image data, edge image data extracted from the video signal of the current frame by the edge image extraction means, and the still image stored in the still edge buffer. 2. A stationary edge detection unit configured to use a portion overlapping with edge image data as the new stationary edge image data and store the new stationary edge image data in the stationary edge buffer. Telop detection device.   The still edge detection means is substantially the same for the edge image data extracted from the video signal of the first frame by the edge image extraction means or the still edge image data by the continuation state detection means over the predetermined number of frames. 2. The telop according to claim 1, wherein the edge image data extracted by the edge image extraction means from the video signal of the frame immediately after being determined to have continued with the content of the image is used as the still edge image data as it is. Detection device.   The continuation state determination means includes an edge number measurement unit that determines whether or not the number of edges indicated by the still edge image data is greater than a first threshold value, and the edge number measurement unit sets the edge number to the first threshold value. When it is determined that the frame count value is greater than 1, the frame count value is increased by 1. When the edge number measurement unit determines that the edge count is less than or equal to the first threshold value, the frame count value is increased to the second threshold value. A duration measurement unit for determining whether or not the frame count value is larger, and a reset for initializing the frame count value to 0 when the edge number measurement unit determines that the number of edges is less than or equal to the first threshold value The telop area is determined by the telop determination unit when the frame count value is larger than the second threshold value. -Up detection device.   The telop determination unit divides the determined still edge image data into blocks each including a predetermined plurality of pixels, and an edge representing a block including a predetermined number of edges or more according to the determined still edge image data An edge block creation unit that creates block image data, a block removal unit that removes an isolated block among blocks indicated by the edge block image data, and an edge block image data after the isolated block is removed by the block removal unit Block number determining means for determining whether or not the number of blocks is greater than a third threshold; and when the block number determining means determines that the number of blocks is greater than the third threshold, after removing the isolated block The smallest rectangular block that encloses the area containing all the blocks indicated by the edge block image data A block dense area for calculating an area ratio between block surrounding area setting means for setting a block area as a block surrounding area and a block portion in the block surrounding area and determining whether the area ratio is equal to or greater than a fourth threshold value; The telop detection device according to claim 1, further comprising: a degree calculation unit, wherein the block surrounding area is determined as the telop area when the area ratio is equal to or greater than the fourth threshold value.   The telop determination means includes an edge removal means for removing an isolated edge from the edges indicated by the determined still edge image data, and the number of edges indicated by the still edge image data after the isolated edge is removed by the edge removal means. Edge number determining means for determining whether the number of edges is greater than a fifth threshold; and a stationary edge after removal of the isolated edge when the edge number determining means determines that the number of edges is greater than the fifth threshold An area ratio between edge surrounding area setting means for setting a minimum rectangular area surrounding an area including all edges indicated by image data as an edge surrounding area and an edge portion in the edge surrounding area is calculated, and the area ratio is the sixth. An edge density calculation unit that determines whether or not the threshold value is greater than or equal to a threshold value, and the edge enclosure region when the area ratio is equal to or greater than the sixth threshold value. Telop detecting device according to claim 1, wherein determining that the telop area. A recorder for recording an input video signal on a recording medium,
Edge image extraction means for extracting edge image data indicating an edge on luminance for each frame from an input video signal or a video signal recorded on the recording medium;
Still edge detection means for detecting, as still edge image data, an overlapping portion between the edge image data extracted from the video signal of the current frame by the edge image extraction means and the edge image data extracted from the video signal of the previous frame at least. When,
Continuation state determination means for determining whether or not the still edge image data has continued with substantially the same content over a predetermined number of frames;
When it is determined by the continuation state detection means that the still edge image data has continued with substantially the same content over the predetermined number of frames, according to the edge density indicated by the determined still edge image data And a telop determination means for determining a telop area.   10. The recorder according to claim 9, further comprising chapter menu creation means for creating a chapter menu using the telop area determined by the telop determination means. Edge image extraction means for extracting edge image data indicating an edge on luminance for each frame from an input video signal;
Edge block creating means for dividing one frame into blocks each composed of a plurality of predetermined pixels and creating edge block image data representing a block including a predetermined number of edges according to the edge image data;
Still block detection means for detecting, as the still block image data, an overlapping portion of the edge block image data of the frame created this time by the edge block creation means and at least the edge block image data of the previously created frame;
Continuation state determining means for determining whether or not the still block image data continues with substantially the same content over a predetermined number of frames;
When it is determined by the continuation state detection means that the still block image data has continued with substantially the same content over the predetermined number of frames, according to the block density indicated by the determined still block image data And a telop determining means for determining a telop area.   The edge image extraction means includes a frame extraction unit for extracting a video signal for one frame of the input video signal, and a luminance difference between the luminance of adjacent pixels from the video signal for one frame extracted by the frame extraction unit is predetermined. 12. The telop detection device according to claim 11, further comprising: an intra-frame edge extraction unit that extracts edge video data indicating edge pixels having a luminance equal to or higher than a value.   The still block detection means includes a still block buffer that stores the still block image data, an edge block image data of a frame that is currently created by the edge block creation means, and the still block image data that is saved in the still block buffer. 12. The telop detection according to claim 11, further comprising: a still block detection unit configured to use a portion overlapping with the new still block image data as new still block image data and store the new still block image data in the still block buffer. apparatus.   The still block detection means includes the edge block image data of the first frame created by the edge block creation means, or the same content of the still block image data over the predetermined number of frames by the continuation state detection means. 12. The telop detection device according to claim 11, wherein the edge block image data of the frame immediately after it is determined that the continuation is continued is used as the still block image data as it is.   The continuation state determining means includes: a block number measuring unit that determines whether or not the number of blocks indicated by the still block image data is greater than a first threshold value; and the block number measuring unit sets the block number to the first threshold value. When it is determined that the number of blocks is larger, the frame count value is increased by 1. When the block number measurement unit determines that the number of blocks is equal to or smaller than the first threshold value, the frame count value is increased to the second threshold value. A duration measurement unit for determining whether or not the number of blocks is greater, and a reset that initializes the frame count value to 0 when the number of blocks measurement unit determines that the number of blocks is equal to or less than the first threshold value. Means for determining a telop area by the telop determination means when the frame count value is larger than a second threshold value. Telop detecting device according to item 11, wherein.   The telop determination unit includes a block removal unit that removes an isolated block among blocks indicated by the edge block image data, and the number of blocks indicated by the edge block image data after removal of the isolated block by the block removal unit is a third threshold value. Block number determining means for determining whether or not the block number is larger, and when the block number determining means determines that the number of blocks is greater than the third threshold, the edge block image data after the isolated block removal is An area ratio between block surrounding area setting means for setting a minimum rectangular block area surrounding the area including all the blocks shown as a block surrounding area and a block portion in the block surrounding area is calculated, and the area ratio is equal to or greater than a fourth threshold value A block density calculation unit that determines whether or not the area ratio is Serial telop detecting apparatus according to claim 11, wherein said block surrounding area when it fourth threshold value or more and judging with the telop area. A recorder for recording an input video signal on a recording medium,
Edge image extraction means for extracting edge image data indicating an edge on luminance for each frame from an input video signal or a video signal recorded on the recording medium;
Edge block creating means for dividing one frame into blocks each composed of a plurality of predetermined pixels and creating edge block image data representing a block including a predetermined number of edges according to the edge image data;
Still block detection means for detecting, as the still block image data, an overlapping portion of the edge block image data of the frame created this time by the edge block creation means and at least the edge block image data of the previously created frame;
Continuation state determining means for determining whether or not the still block image data continues with substantially the same content over a predetermined number of frames;
When it is determined by the continuation state detection means that the still block image data has continued with substantially the same content over the predetermined number of frames, according to the block density indicated by the determined still block image data A telop determination means for determining a telop area,
A recorder comprising:   18. The recorder according to claim 17, further comprising chapter menu creation means for creating a chapter menu using the telop area determined by the telop determination means. Extract edge image data indicating the edge on the brightness for each frame from the input video signal,
Detecting the overlapping portion of the edge image data extracted from the video signal of the current frame and at least the edge image data extracted from the video signal of the previous frame as the still edge image data;
It is determined whether the still edge image data continues with almost the same content over a predetermined number of frames,
When it is determined that the still edge image data continues with substantially the same content over the predetermined number of frames, the telop area is determined according to the density of edges indicated by the determined still edge image data. A telop detection method characterized by the above. Extract edge image data indicating the edge on the brightness for each frame from the input video signal,
One frame is divided into blocks each composed of a plurality of predetermined pixels, and edge block image data representing a block including a predetermined number of edges or more according to the edge image data is created,
Detecting an overlapping portion of the edge block image data of the frame created this time and at least the edge block image data of the previously created frame as the still block image data,
It is determined whether the still block image data continues with almost the same content over a predetermined number of frames,
When it is determined that the still block image data continues with substantially the same content over the predetermined number of frames, the telop area is determined according to the block density indicated by the determined still block image data. A telop detection method characterized by the above.

Images