JP4967798B2 - Detection apparatus and method, program, and recording medium - Google Patents

Description

  The present invention relates to a detection apparatus and method, a program, and a recording medium, and in particular, a detection apparatus and method, a program, and a program that can realize edge detection that is robust to the sharpness characteristics of an input image with a smaller amount of calculation. The present invention relates to a recording medium.

  In recent years, high definition and high definition (HD (high definition)) of broadcast image signals are progressing, but standard definition (SD (Standard Definition)) images are still converted (up-converted) to HD. Broadcasting of image signals is also performed.

  In addition, there are cases where HD telops and SD telops are mixed for materials recorded as HD images.

  In recent years, with the widespread use of HDD (Hard Disk Drive) recorders and the like, there has been a growing need to easily search for specific scenes by analyzing image signals, especially images containing telops. The need to detect is increasing.

  A technique that enables up-conversion from an SD signal with improved resolution to an HD signal has also been proposed (see, for example, Patent Document 1).

Japanese Patent No. 3326879

  Conventionally, telop detection is performed by detecting edges in an image and determining whether or not a telop exists based on the number of detected edges.

  However, in an image in which SD telop and HD telop are mixed, SD telop whose outline is relatively ambiguous is detected from HD material which is an image with high sharpness and whose outline is emphasized. Therefore, the conventional method cannot detect the telop properly.

  The present invention has been made in view of such a situation, and makes it possible to realize edge detection that is robust to the sharpness characteristics of an input image with a smaller amount of calculation.

  One aspect of the present invention is a detection device that detects information for specifying a region including a telop from input image data, and the image data is divided into N pixels in a horizontal direction and a vertical direction, respectively. Dividing means for dividing the plurality of sub-blocks composed of M pixels into a plurality of sub-blocks, and for each of the plurality of sub-blocks, an average value of pixel values of N pixels in the horizontal direction constituting the sub-blocks. Horizontal average value calculating means for calculating an average value in M horizontal directions, and an average value of pixel values of M pixels in the vertical direction constituting the sub-block for each of the plurality of sub-blocks. A vertical average value calculating means for calculating an average value in the N vertical directions, and a difference between a maximum value and a minimum value among the M average values in the horizontal direction, and an area including the telop Identify A horizontal index calculation unit that calculates a first index for the first index, and a second index for specifying a region that includes the telop, using a difference between a maximum value and a minimum value among the N average values in the vertical direction. And a vertical index calculation unit that calculates the index.

The first index is extracted for a subblock of interest and a plurality of subblocks adjacent to the subblock, and data in which the first index is arranged at a position corresponding to the position of each subblock, Horizontal processing data obtained by processing the first index of each of the plurality of sub-blocks by a horizontal filter that multiplies a preset numerical value, the sub-block of interest, and adjacent to the sub-block a plurality of sub-blocks by extracting the second index for the second index, the data are arranged in positions corresponding to the positions of the respective sub-blocks, each of said second plurality of sub-blocks Based on the vertical processing data obtained by processing each of the indicators by a vertical filter that multiplies a preset numerical value, A telop information output means for determining whether or not a telop is included in the sub-block to be noted, and outputting information indicating the determination result as a third index for specifying an area including the telop; Further, it can be provided.

  The telop information output means further generates oblique processing data based on the horizontal processing data and the vertical processing data, the value of the horizontal processing data exceeds a first threshold value, and the value of the vertical processing data Exceeds the second threshold value and the value of the oblique processing data exceeds the third threshold value, it can be determined that the target sub-block contains a telop.

In the horizontal filter, when the image of the stripe pattern in the horizontal direction is included in the sub-block of interest and a plurality of sub-blocks adjacent to the sub-block, the horizontal processing data having a high value is obtained. When a numerical value is set and the vertical filter includes an image of a stripe pattern in the vertical direction in the target sub-block and a plurality of sub-blocks adjacent to the sub-block, a high value of the vertical processing data is obtained. As described above, the numerical value can be set.

  One aspect of the present invention is a detection method of a detection device that detects information for specifying a region including a telop from input image data, and the image data is divided into N pixels in the horizontal direction. , Divided into a plurality of sub-blocks composed of M pixels in the vertical direction, and for each of the plurality of sub-blocks, an average value of pixel values of N pixels in the horizontal direction constituting the sub-block. An average value of M horizontal directions is calculated, and for each of the plurality of sub-blocks, an average value of pixel values of M pixels in the vertical direction constituting the sub-block, and N And calculating a difference between the maximum value and the minimum value among the M average values in the horizontal direction as a first index for specifying a region including the telop, N vertical directions The difference between the maximum value and the minimum value of the average value, is a detection method comprising the steps of calculating a second index for identifying a region including the telop.

  One aspect of the present invention is a program for causing a computer to perform detection processing for detecting information for specifying a region including a telop from input image data, the computer performing horizontal processing on the image data. Dividing means for dividing into a plurality of sub-blocks composed of N pixels in the direction and M pixels in the vertical direction, and for each of the plurality of sub-blocks, each of N in the horizontal direction constituting the sub-block Horizontal average value calculating means for calculating M average values in the horizontal direction, and for each of the plurality of sub-blocks, M in the vertical direction constituting each of the sub-blocks. A vertical average value calculating means for calculating an average value in N vertical directions, and a maximum value among the M horizontal average values; A horizontal index calculation unit that calculates a difference between the minimum values as a first index for specifying a region including the telop; and a difference between the maximum value and the minimum value among the N average values in the vertical direction. , A program that functions as a vertical index calculation unit that calculates a second index for specifying a region including the telop.

  In one aspect of the present invention, the image data is divided into a plurality of sub-blocks each composed of N pixels in the horizontal direction and M pixels in the vertical direction, and for each of the plurality of sub-blocks, An average value of pixel values of each of N pixels in the horizontal direction constituting the sub-block, and an average value in M horizontal directions is calculated, and each of the plurality of sub-blocks constitutes the sub-block. An average value of pixel values of each of M pixels in the vertical direction, N average values in the vertical direction are calculated, and a difference between the maximum value and the minimum value among the M horizontal average values is calculated. , Calculated as a first index for specifying the region including the telop, and the difference between the maximum value and the minimum value among the N average values in the vertical direction specifies the region including the telop. As a second indicator for It is calculated.

  According to the present invention, edge detection that is robust to the sharpness characteristic of an input image can be realized with a smaller amount of calculation.

  Embodiments of the present invention will be described below. Correspondences between constituent elements of the present invention and the embodiments described in the specification or the drawings are exemplified as follows. This description is intended to confirm that the embodiments supporting the present invention are described in the specification or the drawings. Therefore, even if there is an embodiment which is described in the specification or the drawings but is not described here as an embodiment corresponding to the constituent elements of the present invention, that is not the case. It does not mean that the form does not correspond to the constituent requirements. Conversely, even if an embodiment is described here as corresponding to a configuration requirement, that means that the embodiment does not correspond to a configuration requirement other than the configuration requirement. It's not something to do.

  A detection device according to one aspect of the present invention is a detection device that detects information for specifying a region including a telop from input image data, and the image data is divided into N pixels in the horizontal direction. Dividing means (for example, the region extracting unit 201 in FIG. 5) that divides a plurality of sub-blocks configured by M pixels in the vertical direction, and the horizontal that constitutes the sub-block for each of the plurality of sub-blocks. Horizontal average value calculating means (for example, the arithmetic unit 221 in FIG. 6) for calculating the average value of M horizontal directions, which is an average value of pixel values of N pixels in each direction, and the plurality of sub-blocks 6 is an average value of pixel values of M pixels in the vertical direction constituting the sub-block, and is a vertical average value calculating means (for example, FIG. 6). Performance 222) and a difference between the maximum value and the minimum value among the M average values in the horizontal direction as a first index (for example, horizontal edge strength) for specifying a region including the telop A horizontal index calculation unit (for example, the calculator 241 in FIG. 6) to be calculated and a difference between the maximum value and the minimum value among the N average values in the vertical direction are used to specify a region including the telop. A vertical index calculation unit (for example, the calculator 242 in FIG. 6) that calculates the second index (for example, vertical edge strength).

The detection apparatus extracts the first index for a target sub-block and a plurality of sub-blocks adjacent to the sub-block, and sets the first index to a position corresponding to the position of each sub-block. A horizontal filter (for example, the horizontal filter 262 in FIG. 7) that multiplies the arranged data (for example, the data h0 in FIG. 7) by a predetermined numerical value for each of the first indexes of the plurality of sub-blocks. ) To extract the second index for the horizontal processing data (for example, the data h1 in FIG. 7) obtained by processing, the target sub-block, and a plurality of sub-blocks adjacent to the sub-block, Data in which the second index is arranged at a position corresponding to the position of each sub-block (for example, data v0 in FIG. 7) is assigned to each of the plurality of sub-blocks. Vertical filter for multiplying to said second index of, respectively, a value set in advance, respectively (e.g., vertical filter 261 in FIG. 7) vertical processing data obtained is treated with (e.g., data v1 in Fig. 7) Based on the above, it is determined whether or not the target sub-block includes a telop, and information indicating the determination result is output as a third index for specifying the region including the telop. A telop information output unit (for example, the telop area detection unit 203 in FIG. 5) can be further provided.

  A detection method according to one aspect of the present invention is a detection method of a detection device that detects information for specifying a region including a telop from input image data, and the image data is respectively N in the horizontal direction. It is divided into a plurality of sub-blocks composed of M pixels and M pixels in the vertical direction (for example, the process of step S101 in FIG. 10), and the horizontal lines that constitute the sub-blocks for each of the plurality of sub-blocks. An average value of pixel values of N pixels in each direction, and an average value in M horizontal directions is calculated (for example, the process of step S121 in FIG. 11). For each of the plurality of sub-blocks, An average value of pixel values of each of M pixels in the vertical direction constituting the sub-block, and an average value in N vertical directions is calculated (for example, processing in step S123 of FIG. 11). A difference between a maximum value and a minimum value among the M average values in the horizontal direction is calculated as a first index (for example, horizontal edge strength) for specifying a region including the telop, and the N The difference between the maximum value and the minimum value among the average values in the vertical direction is calculated as a second index (for example, vertical edge strength) for specifying the region including the telop (for example, FIG. 11). Steps S125 and S126).

  Embodiments of the present invention will be described below with reference to the drawings.

  First, conventional telop detection processing will be described.

  When detecting a telop included in an image, whether or not a telop is included in the area based on information indicating the edge strength of the detected edge, etc. A method of determining whether or not is used. This is because a telop that is a character is usually composed of a combination of strokes, and for example, an edge often appears repeatedly like a striped pattern image.

  FIG. 1 is a diagram for explaining an example of edge detection that is conventionally performed to detect a telop. In the figure, a Sobel filter is used.

  In FIG. 1, an image for detecting a telop is divided into regions called sub-blocks composed of a plurality of pixels in the horizontal direction and the vertical direction, and is input to a Sobel filter. In this example, a sub-block composed of a predetermined number of pixels arranged in a rectangular area is input to the Sobel filter.

  The Sobel filter is configured to include a horizontal filter 21 and a vertical filter 22, and numerical values described in the respective filters are set as pixel values of a target pixel in the input sub-block and a plurality of pixels adjacent to the target pixel. Are multiplied by the multiplier 31 and the multiplier 32, respectively. In this example, each of the horizontal filter 21 and the vertical filter 22 is configured to correspond to eight pixels adjacent to the target pixel, with the center pixel being the target pixel. Such processing by the horizontal filter 21 and the vertical filter 22, and the multiplier 31 and the multiplier 32 is performed on each pixel of the sub-block.

  Then, threshold determination processing by the threshold determination unit 41 and the threshold determination unit 42 is performed on each of the sub-block pixels that have undergone the processing by the horizontal filter 21 and the vertical filter 22, and the multiplier 31 and the multiplier 32. As a result, for each pixel of the sub-block, a pixel considered to constitute an edge in each of the horizontal direction and the vertical direction is detected.

  Further, the totalizer 51 and the totalizer 52 detect the total value of each pixel considered to constitute an edge in each of the horizontal direction and the vertical direction, and the total value is determined based on the horizontal edge strength and the vertical direction. Is output as the edge strength.

  Based on the edge strength in the horizontal direction and the edge strength in the vertical direction thus obtained, it is determined whether or not a telop is included in the sub-block.

  However, the method described above with reference to FIG. 1 may not accurately detect whether a telop is included.

  FIG. 2 is a diagram illustrating an example of an image including a telop. In the image 100 shown in the figure, telops are displayed in the areas 101 and 102, respectively. The telop displayed in the area 101 is a telop displayed as a high definition (HD) image. The telop displayed in the area 102 is a telop displayed as an image having a standard resolution (SD (Standard Definition)) lower than that of the HD image.

  That is, the image 100 is an image in which an HD image and an SD image are mixed, and such an image is, for example, an image obtained by editing a telop on a material recorded as an HD image, or an SD In recent years, many of these images have been converted to HD (up-converted).

  In an image in which SD telops and HD telops are mixed, SD image telops with relatively vague outlines are detected from HD image materials with high sharpness and enhanced outlines. Therefore, the conventional method cannot detect the telop properly.

  That is, when detecting a pixel considered to constitute an edge in each of the horizontal direction and the vertical direction for each pixel of the sub-block in an HD image that is an image with high sharpness, the threshold value determination unit 41 and the threshold value in FIG. Unless a threshold value that is high to some extent is set in the determination unit 42, pixels that are not related to the edge are detected as pixels constituting the edge. In addition, in the SD image that is an image with low sharpness, when detecting pixels that are considered to constitute edges in the horizontal direction and the vertical direction for each pixel of the sub-block, the threshold value determination unit 41 and the threshold value in FIG. If a threshold value that is low to some extent is not set in the determination unit 42, the number of pixels that are not detected although they are pixels constituting the edge increases.

  FIGS. 3 and 4 are diagrams for explaining problems in the case of performing edge detection of an image in which an HD image and an SD image are mixed according to the conventional method described above with reference to FIG.

  FIG. 3 shows an example in which the edge strength in the vertical direction is detected by a conventional method, for example, for a sub-block 141 of a part of the HD image in the region 101 and a sub-block 142 of a part of the SD image in the region 102. FIG. 6 is a diagram illustrating an example when a threshold suitable for an HD image is set in the threshold determination unit 42.

  In the sub-block 141, for example, a part of the characters is displayed by the pixels in the 10th to 18th columns counting from the left in the figure, and the pixels in the 1st to 9th columns are The background of the character is displayed. The sub-block 141 is an HD image, and some characters are clearly displayed in black and the background is clearly displayed in white.

  On the other hand, the sub-block 142 is an SD image, and a boundary (edge) between a part of a character and the background is displayed in an ambiguous manner. In this example, each pixel in the seventh to twelfth columns from the left in the figure is displayed as a color that is neither black nor white (hatching in the figure).

  When the threshold determination processing is performed on the sub-block 141 and the sub-block 142 by the threshold determination unit 42 that performs the processing by the vertical filter 22 and the multiplier 32 described above and sets a threshold suitable for the HD image, On the right side of the vertical filter 22 in FIG. 3, pixels displayed as black squares in the drawing are detected as pixels that are considered to constitute edges in the vertical direction.

  In the case of FIG. 3, since a threshold suitable for an HD image is set in the threshold determination unit 42, in the sub-block 141, pixels corresponding to the boundary between a part of characters and the background, the ninth column and the The pixels in the tenth column are all detected as pixels constituting the edge. As a result, the vertical edge strength of the sub-block 141 is “8”, which is the total value of the number of pixels in the ninth and tenth columns.

  However, in the sub-block 142 of the SD image, each pixel in the seventh column to the twelfth column gradually changes from white to black, so that the threshold determination unit 42 in which a threshold suitable for the HD image is set. In this threshold determination, there is no pixel detected as a pixel constituting the edge. As a result, the edge strength in the vertical direction of the sub-block 142 becomes “0”.

  FIG. 4 shows an example in which vertical edge strength is detected by using a conventional method, for example, for a sub-block 141 of a part of the HD image in the region 101 and a sub-block 142 of a part of the SD image in the region 102. It is a figure which shows the example at the time of setting the threshold value suitable for SD image to the threshold determination part 42. FIG.

  As in the case of FIG. 3, threshold determination by the threshold determination unit 42 that performs processing by the vertical filter 22 and the multiplier 32 described above on the sub-block 141 and the sub-block 142 and sets a threshold suitable for the SD image. When the above process is performed, pixels displayed as black squares in the drawing on the right side of the vertical filter 22 in FIG. 4 are detected as pixels that are considered to form edges in the vertical direction.

  In the case of FIG. 4, since a threshold suitable for the SD image is set in the threshold determination unit 42, in the sub-block 142, pixels corresponding to the boundary between a part of characters and the background, the seventh column to the seventh column The pixels in the 12th column are all detected as pixels constituting the edge. As a result, the vertical edge strength of the sub-block 142 is “24”, which is the total value of the number of pixels in the seventh to twelfth columns.

  However, in the HD image sub-block 141, the color of the pixel changes only between the ninth column and the tenth column. Therefore, in the threshold determination of the threshold determination unit 42, the pixel is detected as a pixel constituting the edge. Only the pixels in the ninth column and the tenth column are detected as pixels constituting the edge. As a result, the edge strength in the vertical direction of the sub-block 141 becomes “8”.

  As described above, in the conventional method, if the threshold value of the threshold determination unit 42 is suitable for an HD image, no edge is detected in the SD image. If the threshold value is suitable for an SD image, In the case of an SD image, the edge strength becomes larger.

  3 and 4, only the edge strength in the vertical direction has been described for the sake of simplicity, but the edge strength in the horizontal direction is also difficult to accurately detect with the conventional method. .

  In addition, since the sharpness of an image is high in an HD image, for example, in an image in which dark-colored pixels are displayed in a granular shape such as stone or sand, these pixels are mistaken as pixels constituting an edge. May be detected.

  For example, a region 103 in FIG. 2 is an image in which a part of a stone wall is displayed, and a fine granular pattern is displayed. Such an image may be erroneously detected as an edge.

  Therefore, in the present invention, it is possible to appropriately detect a telop even in an image in which an SD image and an HD image are mixed, and further, an image on which a fine grain pattern or the like is displayed is not erroneously detected as an edge.

  FIG. 5 is a block diagram illustrating a configuration example of the telop detection device according to the embodiment of the present invention. In this example, the telop detection device 200 includes a region extraction unit 201, a sub-block edge detection unit 202, a telop region detection unit 203, and a control unit 204. For example, the region extraction unit 201 through the control by a bus or the like. The parts 204 are connected to each other.

  In the figure, an area extraction unit 201 extracts, for example, an area for detecting a telop in an image of image data that has been accepted via the control unit 204, and further, the area is suitable for telop detection. Divide into sub-blocks.

  For example, the sub-block edge detection unit 202 receives the image data of the sub-blocks divided by the region extraction unit 201 via the control unit 204, and determines the horizontal and vertical edge strengths of the sub-blocks, respectively. To detect.

  The telop area detection unit 203 acquires data including a plurality of subblocks, each of which has detected edge strengths in the horizontal direction and the vertical direction by the subblock edge detection unit 202, via the control unit 204, for example. It is determined whether or not each of the sub-blocks is a sub-block including a telop, and information relating to an area composed of sub-blocks determined to be a sub-block including a telop is output.

  The control unit 204 includes, for example, a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), and the like. For example, an input is accepted via an operation unit (not shown). Each part of the telop detection device 200 is controlled based on a user instruction, software such as a program stored in a ROM or the like in advance.

  FIG. 6 is a diagram illustrating a detailed configuration example of the sub-block edge detection unit 202 of FIG. Here, a detailed configuration of the sub-block edge detection unit 202 is described by taking as an example a case where image data including 32 pixels of 8 pixels in the horizontal direction and 4 pixels in the vertical direction is supplied as the sub-block 220 as the sub-block. Will be described.

  As shown in the figure, the sub-block edge detecting unit 202 has pixel values of eight pixels in the first to fourth rows from the top in the horizontal direction of the supplied sub-block 220 in the drawing. The average value is calculated by the calculator 221. That is, the arithmetic unit 221 causes the average value of the eight pixel values in the first row of the sub-block 220, the average value of the eight pixel values in the second row, and the eight pixel values in the third row. The average value of the pixel values and the average value of the pixel values of the eight pixels in the fourth row are respectively calculated. The value calculated here is referred to as four average values in the horizontal direction.

  In addition, the sub-block edge detection unit 202 calculates the average value of the pixel values of the four pixels in the first to eighth columns from the left in the vertical direction of the supplied sub-block 220 in the calculator 222. To calculate each. That is, the calculator 222 calculates the average value of the four pixels in the first column of the sub-block 220, the average value of the pixel values of the four pixels in the second column, and the four pixel values in the third column. The average value of the pixel values,..., And the average value of the pixel values of the four pixels in the eighth column are calculated. The value calculated here is referred to as eight average values in the vertical direction.

  The sub-block edge detection unit 202 extracts the minimum value among the four average values in the horizontal direction by the minimum value extraction unit 231, and calculates the maximum value among the four average values in the horizontal direction as the maximum value extraction unit 232. Extract by In addition, the sub-block edge detection unit 202 extracts the minimum value of the eight average values in the vertical direction by the minimum value extraction unit 233, and extracts the maximum value of the eight average values in the vertical direction as the maximum value. Extracted by the unit 234.

  Then, the sub-block edge detection unit 202 calculates the difference between the minimum value extracted by the minimum value extraction unit 231 and the maximum value extracted by the maximum value extraction unit 232 by the calculator 241, and the value obtained as a result Is output as the edge strength in the horizontal direction. Similarly, the sub-block edge detection unit 202 calculates the difference between the minimum value extracted by the minimum value extraction unit 233 and the maximum value extracted by the maximum value extraction unit 234 by the calculator 242 and obtained as a result. Outputs the value as the vertical edge strength.

  In this way, a plurality of average values in the horizontal direction and the vertical direction of each pixel constituting the sub block are calculated, and the difference between the maximum value and the minimum value is set as the edge strength in the horizontal direction and the vertical direction, respectively. Thus, for example, even in the case of an ambiguous image that gradually changes from white to black as in the case of the sub-block 142 in FIGS. 3 and 4, an appropriate edge strength can be calculated. As in the case of the sub-block 141, an appropriate edge strength can be calculated even for an image that sharply changes from white to black (high sharpness). Therefore, according to the sub-block edge detection unit 202, it is possible to detect substantially the same edge strength in the sub-block of the edge portion regardless of whether it is an SD image or an HD image.

  Furthermore, by calculating a plurality of average values in the horizontal direction and vertical direction of each pixel constituting the sub-block, the difference between the maximum value and the minimum value as the edge strength in the horizontal direction and vertical direction, respectively, For example, it is possible to prevent an image or the like on which a fine granular pattern is displayed as shown in the area 103 of FIG. 2 from being erroneously detected as an edge. For example, in a sub-block of an image in which black dots are regularly scattered on a white background, all pixels in one row (or all pixels in one column) in the horizontal direction (or vertical direction) of each pixel When the average value of the values is calculated, the value is the same as the pixel value of the gray pixel in any row (or column), and the difference between the maximum value and the minimum value of the plurality of average values is almost “0”. Because.

  In addition, by doing so, it is also possible to suppress the calculation amount. For example, in the sobel filter described above with reference to FIG. 1, by passing through the horizontal filter 21 and the vertical filter 22, an operation is performed at a position where the value in the horizontal filter 21 and the vertical filter 22 is not 0. Although a total of 12 (= 6 + 6) multiplications are required for each pixel, a plurality of average values in the horizontal direction and the vertical direction of each pixel are calculated, and the difference between the maximum value and the minimum value is calculated. By setting the edge strength in the horizontal direction and the vertical direction, it is possible to reduce the amount of calculation performed for each pixel.

  Here, an example has been described in which image data composed of 32 pixels of 8 pixels in the horizontal direction and 4 pixels in the vertical direction is supplied as a sub-block. However, in the case of detecting edges in order to detect telops The number of pixels in the horizontal and vertical directions of the sub-block is preferably set to the number of pixels corresponding to the thickness of the telop character stroke to be detected.

  Further, a part of the sub-blocks may be overlapped (overlapped). For example, the subblock may be divided so that some of the plurality of pixels constituting the nth subblock are used as pixels of the (n + 1) th subblock.

  FIG. 7 is a diagram illustrating a detailed configuration example of the telop area detection unit 203 in FIG. 5.

  As described above, the telop area detection unit 203 receives data including a plurality of sub-blocks in which the horizontal and vertical edge intensities are detected by the sub-block edge detection unit 202 via, for example, the control unit 204. get. In this example, data v0 and data h0 composed of nine sub-blocks, three in the horizontal direction and three in the vertical direction, are acquired.

  The data v0 is acquired, for example, as data of a 3 × 3 matrix in which only the value of the vertical edge strength of each subblock is embedded at the position of each subblock, and the data h0 is, for example, Is obtained as matrix data of 3 rows and 3 columns in which only the values of the edge strengths in the horizontal direction of the respective sub blocks are embedded at the positions of the sub blocks.

  The telop area detection unit 203 is configured to include a vertical filter 261 and a horizontal filter 262, and numerical values described in the vertical filter 261 and the horizontal filter 262 are respectively added to the values of the input data v0 and data h0. 271 and a multiplier 272 are respectively multiplied.

  For example, the three values in the first column from the left in the figure of the data v0 are respectively multiplied by “1, 1, 1” that are the values in the first column from the left in the figure of the vertical filter 261, respectively. The three values in the second column of v0 are respectively multiplied by “2, 2, 2”, which are the values in the second column of the vertical filter 261, and the three values in the third column of the data v0 are respectively multiplied. Are respectively multiplied by the values in the third column of the vertical filter 261, “1, 1, 1”. Further, the multiplier 271 calculates a total value of nine values multiplied by the respective numerical values as a vertical edge index.

  The vertical edge index is calculated as a vertical edge index corresponding to the sub-block in the second row and the second column at the center position of the data v0.

  The three values in the first row from the top of the data h0 in the figure are respectively multiplied by “1, 1, 1” which is the value in the first row from the top in the drawing of the horizontal filter 262, respectively. The three values in the second column of h0 are respectively multiplied by “2, 2, 2”, which are the values in the second row of the horizontal filter 262, and the three values in the third row of the data h0 are respectively multiplied. Are respectively multiplied by “1, 1, 1” which is the value in the third row of the horizontal filter 262. Further, the multiplier 272 calculates a total value of nine values multiplied by the respective numerical values as a horizontal edge index.

  This horizontal edge index is also calculated as a horizontal edge index corresponding to the second row and second column sub-block, which is also the center position of the data h0.

  In the telop area detection unit 203, the vertical edge index is supplied to the threshold value determination unit 281 as data v1. The threshold determination unit 281 determines whether or not the value of the data v1 is greater than a preset vertical threshold. For example, when it is determined that the value of the data v1 is greater than a preset vertical threshold, the value “1” is set. ”Is output, and when it is determined that the value of the data v1 is not greater than the vertical threshold, the value“ 0 ”is output.

  In the telop area detection unit 203, the vertical edge index is supplied to the threshold value determination unit 283 as data h1. The threshold value determination unit 283 determines whether or not the value of the data h1 is greater than a preset horizontal threshold value. For example, when it is determined that the value of the data h1 is greater than a preset horizontal threshold value, the value “1 ”Is output, and when it is determined that the value of the data h1 is not greater than a preset lateral threshold value, the value“ 0 ”is output.

  Further, in the telop area detection unit 203, the data v1 and the data h1 are added by the adder 273 and supplied to the threshold value determination unit 282 as data d1. The data d1 represents the oblique edge index. The threshold determination unit 282 determines whether or not the value of the data d1 is larger than a preset oblique threshold. For example, when it is determined that the value of the data d1 is larger than a preset oblique threshold, the value “1” is determined. ”Is output, and when it is determined that the value of the data d1 is not larger than a preset oblique threshold value, the value“ 0 ”is output.

  Then, the logical product calculator 291 calculates and outputs the logical product of the output values of the threshold value determination unit 281 to the threshold value determination unit 283. For example, when the operation result of the AND operator 291 is a value “1”, the sub-block of interest (the sub-block in the second row and second column of the data v0 (h0)) is assumed to be a sub-block of a telop. The detection result of the telop area detection unit 203 is output. When the operation result of the AND operator 291 is “0”, it is determined that the sub-block of interest (the sub-block in the second row and second column of the data v0 (h0)) is not a telop sub-block. The

  Note that the vertical filter 261 and the horizontal filter 262 are set so that the vertical edge index or the horizontal edge index is high for the sub-blocks including the vertical stripe pattern and the horizontal stripe pattern, respectively. For example, the vertical edge index or the horizontal edge index is set not to be a high value for a sub-block including a vertical single line and a horizontal single line. This is because the telop often constitutes a striped pattern image.

  For example, consider a telop on which characters “river” are displayed as shown in FIG. The character “river” shown in the figure is displayed as a black-lined encircled character (outlined character). When the change in the luminance of the pixel is examined as indicated by an arrow 311 in the center of the figure, The graph is as shown in FIG.

  That is, the brightness is extremely high in the white portion of the character, and the brightness is extremely low in the black line portion of the character. In the background portion of the character, the luminance is about halfway between the white portion and the black line portion. As described above, in the vicinity of the character “river”, the luminance value of the pixel changes 12 times, and the same luminance change characteristic as in the stripe pattern image is seen.

  On the other hand, the above-described characteristics are not seen in an image such as an object that is not a telop. FIG. 9A shows an example of an image in which a dark-colored object is displayed in approximately the right half of the drawing. In such an image, as in the case of FIG. 8A, when the change in the luminance of the pixel is examined as indicated by the arrow 312 near the center in the figure, the graph shown in FIG. 9B is obtained. become.

  That is, the luminance is high in the portion where the background of the object is displayed, and the luminance is low in the portion where the object is displayed. In this way, the brightness value of the pixel changes once in the periphery of an image on which an object with different brightness is displayed instead of a telop, and the same brightness change characteristic as that of a single line image can be seen. It will be.

  For example, the vertical filter 261 has a high vertical edge index of a sub-block having luminance change characteristics as shown in FIG. 8B in itself (the sub-block of interest) and its neighboring sub-blocks. For example, in the sub-block around itself and its surroundings, for example, the vertical edge index of a sub-block having luminance characteristics as shown in FIG. Is set.

  The threshold determination unit 281 also outputs the value “1” only when the vertical edge index of the sub-block is a high value.

  Here, the vertical stripe pattern has been described as an example, but in the region where characters are displayed in the image, the same characteristics as in the stripe pattern image are also seen in the horizontal direction. Accordingly, the horizontal filter 262 is also set so that the horizontal edge index of the sub-block having the same luminance change characteristic as that of the stripe pattern image becomes a high value, as if the luminance is the same as that of the single line image. Is set so that the horizontal edge index of the sub-block having the change characteristic is not a high value, and the threshold determination unit 283 also sets the value “1” only when the horizontal edge index of the sub-block is a high value. It is designed to output.

  Similarly, the threshold determination unit 282 also sets the value “1” only when the sum value (data d1) of the vertical edge index (data v1) and the horizontal edge index (data h1) of the sub-block is high. It is designed to output.

  Note that the vertical filter 261 and the horizontal filter 262 are set so that the vertical edge index or the horizontal edge index is high for each sub-block including a vertical stripe pattern or a horizontal stripe pattern. As long as the vertical edge index or the horizontal edge index does not become a high value for the sub-block including the main line or the horizontal single line, the setting is different from that shown in FIG. You may do it.

  Next, telop detection processing by the telop detection apparatus 200 will be described with reference to the flowchart of FIG.

  In step S101, the region extraction unit 201 divides the image into sub-blocks. At this time, for example, the region extraction unit 201 extracts a region where a telop is to be detected in an image of image data that has been accepted via the control unit 204, and further extracts the region as a subtitle suitable for telop detection. Divide into blocks.

  In step S102, the sub-block edge detection unit 202 performs a sub-block edge detection process described later with reference to FIG. 11 for each sub-block divided in the process of step S201. Thereby, the horizontal edge strength and the vertical edge strength of each sub-block are calculated.

  Here, an example of details of the sub-block edge detection process in step S102 of FIG. 10 will be described with reference to the flowchart of FIG.

  In step S <b> 121, the sub-block edge detection unit 202 calculates the average value of the pixel values of each row in the horizontal direction of the supplied sub-block by the calculator 221.

  In step S122, the sub-block edge detection unit 202 obtains a maximum value and a minimum value for each of the average values of the pixel values in each horizontal row calculated in the process of step S121.

  At this time, for example, as described above with reference to FIG. 6, the average value of the pixel values of the eight pixels in the first to fourth rows in the horizontal direction of the supplied sub-block 220 is calculated. The minimum value of the four average values in the horizontal direction is extracted by the minimum value extraction unit 231, and the maximum value of the four average values in the horizontal direction is calculated as the maximum value extraction unit 232. Extracted by

  In step S123, the sub-block edge detecting unit 202 calculates the average value of the pixel values of each column in the vertical direction of the supplied sub-block by the calculator 222.

  In step S124, the sub-block edge detection unit 202 obtains a maximum value and a minimum value for each of the average values of the pixel values in each column in the vertical direction calculated in the process of step S123.

  At this time, for example, as described above with reference to FIG. 6, the average value of the pixel values of the four pixels in the first to eighth columns in the vertical direction of the supplied sub-block 220 is calculated. The minimum value of the eight average values in the vertical direction is extracted by the minimum value extraction unit 233, and the maximum value of the eight average values in the vertical direction is calculated as the maximum value extraction unit 234. Extracted by

  In step S125, the sub-block edge detecting unit 202 obtains the difference between the maximum value and the minimum value obtained in the process in step S122 and the process in step S124, respectively.

  At this time, for example, as described above with reference to FIG. 6, a difference between the minimum value extracted by the minimum value extraction unit 231 and the maximum value extracted by the maximum value extraction unit 232 is calculated by the calculator 241. The calculator 242 calculates the difference between the minimum value extracted by the minimum value extraction unit 233 and the maximum value extracted by the maximum value extraction unit 234.

  In step S126, the sub-block edge detection unit 202 outputs the value obtained by the process in step S125 as the horizontal edge strength or the vertical edge strength of the sub-block.

  At this time, for example, as described above with reference to FIG. 6, the value of the calculation result of the calculator 241 is the edge strength in the horizontal direction, and the value of the calculation result of the calculator 242 is the edge strength in the vertical direction. Are output in association with information indicating the position of the sub-block 220.

  Note that the above-described steps S121 to S126 are executed for each of the plurality of sub-blocks supplied to the sub-block edge detecting unit 202, and the horizontal edge strength or the vertical direction of each of the plurality of sub-blocks is performed. Edge strength is output.

  In this way, the edge strength of the sub-block is detected (calculated).

  Returning to FIG. 10, after step S <b> 102, in step S <b> 103, the telop area detection unit 203 executes telop area detection processing described later with reference to FIG. 12 based on the processing result of step S <b> 102. As a result, information on the sub-block of the telop is output.

  Here, an example of details of the telop area detection processing in step S103 in FIG. 10 will be described with reference to the flowchart in FIG.

  In step S151, the telop area detection unit 203 extracts the edge strength in the horizontal direction of the plurality of sub-blocks, and in step S152, extracts the edge strength in the vertical direction of the plurality of sub-blocks.

  As a result, for example, as described above with reference to FIG. 7, data v0 is acquired as matrix data in which only the value of the vertical edge strength of each sub-block is embedded at the position of each sub-block. The data h0 is obtained as matrix data in which only the edge strength values in the horizontal direction of the respective sub-blocks are embedded at the positions of the respective sub-blocks.

  In step S153, the telop area detection unit 203 performs filter processing.

  At this time, for example, as described above with reference to FIG. 7, the numerical values described in the vertical filter 261 and the horizontal filter 262 are respectively added to the values of the input data v0 and data h0. The multiplier 271 and the multiplier 272 calculate the total value of the nine values multiplied by the respective numerical values 272 as the vertical edge index and the horizontal edge index.

  The horizontal edge index and the vertical edge index are calculated as the vertical edge index (data v1) and the horizontal edge index (data h1) corresponding to the sub-block at the center position of the data v0 or data h0. .

  In step S154, the telop area detection unit 203 detects vertical, horizontal, and diagonal stripes.

  At this time, for example, as described above with reference to FIG. 7, the threshold value determination unit 281 determines whether the value of the data v1 is larger than a preset vertical threshold value. When it is determined that the value is larger than the set vertical threshold, the value “1” is output, and when it is determined that the value of the data v1 is not larger than the vertical threshold, the value “0” is output. Similarly, the threshold value determination unit 283 compares the value of the data h1 with a preset horizontal threshold value, and outputs a value “1” or a value “0”. Further, the threshold value determination unit 282 compares the value of the data d1 with a preset oblique threshold value, and outputs a value “1” or a value “0”.

  In step S155, the telop area detection unit 203 determines whether a stripe is detected in each of the vertical, horizontal, and diagonal directions. For example, when the logical product calculator 291 in FIG. 7 calculates the logical product of the output values of the threshold value determination unit 281 to the threshold value determination unit 283 and the calculation result is the value “1”, each of vertical, horizontal, and diagonal It is determined that a stripe has been detected in each direction, and the process proceeds to step S156.

  In step S156, the telop area detection unit 203 detects the subblock as being a subblock of the telop.

  On the other hand, when the calculation result of the logical product calculator 291 in FIG. 7 is “0” in step S155, it is determined that no stripe is detected in each of the vertical, horizontal, and diagonal directions, and the process of step S156 is performed. Skipped and processing ends.

  In this way, the sub-block of the telop is detected.

  A telop detection result by the telop detection apparatus 200 of the present invention will be described with reference to FIGS.

  FIG. 13 is a graph for explaining the distribution of the feature amount of the telop image displayed in the image. In the figure, the horizontal axis direction (x-axis direction) in the figure represents the value of the data v1 described above with reference to FIG. 7, and the vertical axis direction (y-axis direction) in the figure refers to FIG. The value of the data h1 described above is represented, and in each of the data v1 and data h1 value sub-blocks, the probability of the presence of a telop is represented in the axial direction (z-axis direction) perpendicular to the paper surface. The numerical values such as “20000” and “40000” shown in the figure are for expressing the high probability of the presence of a telop as a contour line.

  In the figure, the height of the paper surface corresponds to a position where the probability of the presence of a telop is about 50%. The probability that a telop exists will be low. Accordingly, the numerical value of the contour line is “0” at the height of the paper surface of FIG.

  In the same figure, the probability that a telop exists is very low as the numerical values of the contour lines are displayed as “−20000”, “−40000”, etc. on the lower left side in the figure. That is, the probability that a telop exists in the sub-block where both the data v1 and the data h1 are low is very low. In this region, both the data v1 and the data h1 are low values, and for example, an edge like a stripe pattern is not detected in the sub-block image in this region.

  Further, the numerical value of the contour line is also about “0” on the lower right side in the figure, and the probability that the telop exists is very low. In the region 351 in the figure, contour lines are not described, but in actuality, it is about “−20000” to “0”. In the region 351, although the value of the data h1 is low, the value of the data v1 is quite high, so that a considerable edge strength should have been detected. However, for example, the area 351 has a high probability of being a sub-block on which a blind image or the like is displayed, and as a result, the probability that a telop is present is extremely low.

  As described above, it is necessary to set the vertical threshold value, the horizontal threshold value, and the oblique threshold value in FIG. 7 so as to remove the region where the probability that the telop exists is extremely low. In the present invention, for example, as indicated by a line 371, a line 372, and a line 373 in FIG. 13, a vertical threshold value, a horizontal threshold value, and a diagonal threshold value are set, respectively. By doing so, it is possible to prevent a sub-block having a very low probability of the presence of a telop from being detected as a sub-block including the telop. In particular, by setting a line 373 (oblique threshold) in addition to the lines 371 and 372, an area with a very low probability that the lower left telop exists in the figure is a detection target sub-block as a sub-block including the telop. Appropriate removal from the block.

  FIG. 14 is a graph comparing the case where a sub-block including a telop is detected by the conventional method with the case where a sub-block including a telop is detected by the telop detection apparatus 200 of the present invention.

  In the figure, the horizontal axis in the figure is the reproduction rate, and the vertical axis in the figure is the precision rate. When a sub-block including a telop is detected by a conventional method, the telop detection device of the present invention The change in the recall rate and the matching rate when a sub-block including a telop is detected by 200 is indicated by lines 401 and 402, respectively.

  Here, the recall rate and the matching rate will be described.

  For example, a set of sub-blocks that are detected as sub-blocks including a telop by the telop detection apparatus 200 and actually include a telop is defined as tp (true-positive). Fp (false-positive) is a set of sub-blocks detected as sub-blocks including, but not actually including a telop.

  In addition, a set of sub-blocks that are not detected as sub-blocks including the telop by the telop detection apparatus 200 even though the telop is actually included are defined as fn (false-negative), and the telop is actually included. A set of sub-blocks that are not detected by the telop detection apparatus 200 as sub-blocks including a telop is defined as tn (true-negative).

  For example, the relationship between tp, fp, fn, and tn in the set of all the sub-blocks constituting one image is as shown in FIG. That is, the sum of fn and tp is a sub-block (correct answer) that actually includes a telop, and the sum of tp and fp is detected by the telop detection device 200 as a sub-block that includes a telop (detection result). ).

  The recall rate represents the proportion of correct answers actually detected, and is obtained by the following equation.

  Reproducibility = tp / (tp + fn)

  The relevance rate represents the proportion of correct answers in the detection result, and is obtained by the following equation.

  Matching rate = tp / (tp + fp)

  From the viewpoint of the detection performance of the telop detection device, it is desirable to realize both the recall rate and the matching rate, but in practice it is difficult to obtain such a detection result.

  For example, in the image shown in FIG. 2, when a sub-block in an area surrounded by an ellipse as shown in FIG. 16 is detected as a sub-block including a telop by the telop detection device 200, the recall is 100%. . However, the region surrounded by the ellipse in FIG. 16 includes images that are not telops.

  In other words, if all sub-blocks are detected regardless of whether or not a telop is included, tn = fn = 0 and the recall is 100%, but the precision is low.

  For example, in the image shown in FIG. 2, when a sub-block in an area surrounded by an ellipse as shown in FIG. 17 is detected as a sub-block including a telop by the telop detection device 200, the relevance rate is 100%. It becomes. However, the area surrounded by the ellipse in FIG. 17 includes only a small part of the telop image, and other areas to be detected remain.

  In other words, in the extreme case, only one sub-block is detected from the entire screen, and if it is correct, the precision is 100%, but the recall is low.

  Here, a blank area or the like existing in the text string of the telop is also considered as a correct answer (the telop is included), and the ratio between the character and the blank area existing in the character string is set to 50 based on the character spacing. The difference between the conventional method and the present invention will be described with respect to the matching rate when a detection result with a recall rate of 50% is obtained.

  Returning to FIG. 14, for example, when a detection result with a recall rate of 50% is obtained, in the conventional method indicated by the line 401, the precision is 68% (0.68), whereas According to the present invention indicated by 402, the precision is 92% (0.92). In addition, when a detection result with a reproduction rate of 50% is obtained, the error rate is 32% (= 1−0.68) in the conventional method shown by the line 401, while the line 402 According to the present invention shown, the misrecognition rate is 8% (= 1-0.92).

  That is, according to the present invention, it can be said that the relevance rate is improved by about 24% compared to the conventional method, and the misrecognition rate is reduced to about 1/4.

  In addition, the precision of the line 402 approaches 1.0 (100%) as the recall rate approaches 0, whereas the precision of the line 401 becomes 0.2 (100%) as the recall ratio approaches 0. Approaching the degree.

  That is, in the conventional method, for example, it is difficult to obtain a high precision rate even at the sacrifice of the reproduction rate, for example, by setting a threshold value strictly, but in the present invention, the reproduction rate is sacrificed to some extent. If so, it is possible to obtain a higher precision.

  The series of processes described above can be executed by hardware, or can be executed by software. When the above-described series of processing is executed by software, a program constituting the software executes various functions by installing a computer incorporated in dedicated hardware or various programs. For example, a general-purpose personal computer 500 as shown in FIG. 18 is installed from a network or a recording medium.

  In FIG. 18, a CPU (Central Processing Unit) 501 executes various processes according to a program stored in a ROM (Read Only Memory) 502 or a program loaded from a storage unit 508 to a RAM (Random Access Memory) 503. To do. The RAM 503 also appropriately stores data necessary for the CPU 501 to execute various processes.

  The CPU 501, ROM 502, and RAM 503 are connected to each other via a bus 504. An input / output interface 505 is also connected to the bus 504.

  The input / output interface 505 includes an input unit 506 including a keyboard and a mouse, a display including a CRT (Cathode Ray Tube) and an LCD (Liquid Crystal display), an output unit 507 including a speaker, a hard disk, and the like. A communication unit 509 including a storage unit 508, a network interface card such as a modem and a LAN card, and the like is connected. A communication unit 509 performs communication processing via a network including the Internet.

  A drive 510 is connected to the input / output interface 505 as necessary, and a removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory is appropriately attached, and a computer program read from them is loaded. It is installed in the storage unit 508 as necessary.

  When the above-described series of processing is executed by software, a program constituting the software is installed from a network such as the Internet or a recording medium such as the removable medium 511.

  The recording medium shown in FIG. 18 is a magnetic disk (including a floppy disk (registered trademark)) on which a program is recorded, which is distributed to distribute the program to the user, separately from the apparatus main body. Removable media consisting of optical disks (including CD-ROM (compact disk-read only memory), DVD (digital versatile disk)), magneto-optical disks (including MD (mini-disk) (registered trademark)), or semiconductor memory It includes not only those configured by 511 but also those configured by a ROM 502 on which a program is recorded, a hard disk included in the storage unit 508, and the like distributed to the user in a state of being incorporated in the apparatus main body in advance.

  The steps of executing the series of processes described above in this specification are performed in parallel or individually even if they are not necessarily processed in time series, as well as processes performed in time series in the order described. It also includes processing.

It is a figure explaining the example of the edge detection conventionally performed in order to detect a telop. It is a figure which shows the example of the image containing the telop. It is a figure explaining the problem in the case of performing the edge detection of the image in which the image of HD and the image of SD are mixed by the conventional system. It is a figure explaining the problem in the case of performing the edge detection of the image in which the image of HD and the image of SD are mixed by the conventional system. It is a block diagram which shows the structural example of the telop detection apparatus which concerns on one embodiment of this invention. It is a figure which shows the detailed structural example of the subblock edge detection part of FIG. It is a figure which shows the detailed structural example of the telop area | region detection part of FIG. It is a figure explaining the characteristic of the change of the brightness | luminance of the image of a telop character. It is a figure explaining the characteristic of the change of the brightness | luminance of the image which is not a telop. It is a flowchart explaining the example of a telop detection process. It is a flowchart explaining a subblock edge detection process. It is a flowchart explaining a telop area | region detection process. It is a graph explaining distribution of the feature-value of the image of the telop currently displayed in the image. 6 is a graph comparing a case where a sub-block including a telop is detected by a conventional method with a case where a sub-block including a telop is detected according to the present invention. It is a figure explaining a relevance rate and a recall rate. It is a figure explaining a relevance rate and a recall rate. It is a figure explaining a relevance rate and a recall rate. And FIG. 16 is a block diagram illustrating a configuration example of a personal computer.

Explanation of symbols

  200 telop detection device, 201 region extraction unit, 202 sub-block edge detection unit, 203 telop region detection unit, 221 arithmetic unit, 222 arithmetic unit, 231 minimum value extraction unit, 232 maximum value extraction unit, 233 minimum value extraction unit, 234 Maximum value extraction unit, 241 arithmetic unit, 242 arithmetic unit, 261 vertical filter, 262 horizontal filter, 271 multiplier, 272 multiplier, 273 adder, 281 threshold determination unit, 282 threshold determination unit, 283 threshold determination unit, 291 logic Product calculator

Claims (7)

A detection device that detects information for specifying a region including a telop from input image data,
Dividing means for dividing the image data into a plurality of sub-blocks each composed of N pixels in the horizontal direction and M pixels in the vertical direction;
For each of the plurality of sub-blocks, horizontal average value calculating means for calculating the average value of M horizontal directions, which is the average value of the N pixels in the horizontal direction constituting the sub-block, ,
For each of the plurality of sub-blocks, a vertical average value calculating unit that calculates an average value of N vertical values of M pixels in the vertical direction constituting the sub-block, and ,
A horizontal index calculation unit that calculates a difference between a maximum value and a minimum value among the M average values in the horizontal direction as a first index for specifying a region including the telop;
A vertical index calculation unit that calculates a difference between a maximum value and a minimum value among the N average values in the vertical direction as a second index for specifying a region including the telop. The first index is extracted for a subblock of interest and a plurality of subblocks adjacent to the subblock, and data in which the first index is arranged at a position corresponding to the position of each subblock, Horizontal processing data obtained by processing the first index of each of the plurality of sub-blocks by a horizontal filter that multiplies a predetermined numerical value;
The second index is extracted for the target subblock and a plurality of subblocks adjacent to the subblock, and data in which the second index is arranged at a position corresponding to the position of each subblock is obtained. , Based on the vertical processing data obtained by processing each of the second index of the plurality of sub-blocks by a vertical filter that multiplies a predetermined numerical value,
Telop information output means for determining whether or not a telop is included in the sub-block of interest, and outputting information indicating the determination result as a third index for specifying an area including the telop The detection device according to claim 1. The telop information output means includes
Further generating oblique processing data based on the horizontal processing data and the vertical processing data;
When the value of the horizontal processing data exceeds a first threshold, the value of the vertical processing data exceeds a second threshold, and the value of the oblique processing data exceeds a third threshold, the target sub-block The detection device according to claim 2, wherein it is determined that a telop is included. In the horizontal filter, when the image of the stripe pattern in the horizontal direction is included in the sub-block of interest and a plurality of sub-blocks adjacent to the sub-block, the horizontal processing data having a high value is obtained. A numerical value is set,
When the vertical filter includes an image of a stripe pattern in the vertical direction in the sub-block of interest and a plurality of sub-blocks adjacent to the sub-block, the vertical filter is configured to obtain the high-value vertical processing data. The detection device according to claim 2, wherein a numerical value is set. A detection method of a detection device for detecting information for specifying a region including a telop from input image data,
The image data is divided into a plurality of sub-blocks each composed of N pixels in the horizontal direction and M pixels in the vertical direction,
For each of the plurality of sub-blocks, an average value of pixel values of N pixels in the horizontal direction constituting the sub-block, and calculating an average value of M horizontal directions,
For each of the plurality of sub-blocks, an average value of pixel values of M pixels in the vertical direction constituting the sub-block, and calculating an average value of N vertical directions,
A difference between the maximum value and the minimum value among the M horizontal average values is calculated as a first index for specifying a region including the telop,
A detection method comprising: calculating a difference between a maximum value and a minimum value among the N average values in the vertical direction as a second index for specifying a region including the telop. A program for causing a computer to perform detection processing for detecting information for specifying a region including a telop from input image data,
Computer
Dividing means for dividing the image data into a plurality of sub-blocks each composed of N pixels in the horizontal direction and M pixels in the vertical direction;
For each of the plurality of sub-blocks, horizontal average value calculating means for calculating the average value of M horizontal directions, which is the average value of the N pixels in the horizontal direction constituting the sub-block, ,
For each of the plurality of sub-blocks, a vertical average value calculating unit that calculates an average value of N vertical values of M pixels in the vertical direction constituting the sub-block, and ,
A horizontal index calculation unit that calculates a difference between a maximum value and a minimum value among the M average values in the horizontal direction as a first index for specifying a region including the telop;
A program that functions as a vertical index calculation unit that calculates a difference between a maximum value and a minimum value among the N average values in the vertical direction as a second index for specifying a region including the telop.   A recording medium on which the program according to claim 5 is recorded.

Images