JP5264330B2 - Route video management system and video management method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a railroad image management system and a railroad image management method, for automatically acquiring a numeric character displayed on an image. <P>SOLUTION: The railroad image management system includes an image acquisition part which acquires a numeric character embedded in an image output from an image reproduction device, and a numeric character recognition part which recognizes the numeric character as a digital value. A numeric character recognition range 20 composed of cells having a plurality of weighted pixels is defined. A numeric character taken into the numeric character recognition range 20 is weighted for each pixel. The weight values of the numeric character is added for each cell. Whether the added value exceeds a predetermined threshold is determined for each cell. A pattern code as a bit string corresponding to the number of cells and including bits determined by the threshold is generated. <P>COPYRIGHT: (C)2010,JPO&amp;INPIT

Description

  The present invention relates to a route video management system and a video management method for automatically recognizing characters embedded in a video frame.

  In the event of a failure in a railway facility on a railway line, in order to make a recovery plan, it is necessary to check the surroundings of the location where the failure occurred, such as the access route to the location or the heavy equipment installation space. It takes a lot of time and effort to identify the location from the huge amount of video data taken. Therefore, it has been desired to introduce a route video management system capable of automatically editing video data for each predetermined section and easily reproducing a video of a place where a failure or the like has occurred.

  Conventionally, for example, a route video management device disclosed in Patent Document 1 below captures a video around a route taken by a test vehicle, recognizes a change in a sign symbol in the video, and associates the sign symbol with a section. By making the list, it is possible to reproduce the video of the place where the failure occurred.

JP 2007-26169 A

  However, the route image management apparatus disclosed in Patent Document 1 can recognize the change of the sign symbol, but cannot automatically recognize the utility pole number displayed in the image. For this reason, there has been a problem that the telephone pole number, which is effective information for specifying the occurrence location of a failure or the like, cannot be automatically taken into the list.

  The present invention has been made in view of the above, and an object of the present invention is to provide a route video management system and a video management method capable of automatically acquiring numbers displayed on a video.

To solve the above problems and achieve the object, route video management system according to the present invention is a recognized route video management system video output from the video reproduction device, a numeric recognition range surrounding the digits A plurality of cells, and the weights of the pixels corresponding to the numbers captured in a number recognition range in which the plurality of cells are composed of a plurality of pixels are summed for each cell, and the summed value is a predetermined value. It is determined for each cell whether or not a threshold value is exceeded, and consists of a code indicating whether each cell that is a part of the numeral recognition range and that represents the characteristic part of the numeral is in a lit state or a non-lit state A number recognizing unit that generates a bit string and recognizes a number attached to a power pole installed along a route on which a train travels is provided.

  According to the route image management system according to the present invention, a number recognition range including a plurality of cells composed of a plurality of pixels is defined, and a number recognition range forming a circumscribed rectangle surrounding the number is defined and captured in the number recognition range. From the code that sums the weight of each pixel corresponding to the number for each cell, determines whether the sum exceeds a predetermined threshold value for each cell, and indicates whether each cell is lit or unlit As a result, the number displayed on the video can be automatically acquired.

  Embodiments of a route image management system according to the present invention will be described below in detail with reference to the drawings. Note that the present invention is not limited to the embodiments.

Embodiment 1 FIG.
FIG. 1 is a diagram illustrating an example of a configuration of a route video management system according to the first embodiment. The route video management system 100 shown in FIG. 1 includes, as main components, a digital VTR 1, an external interface 2, an input device 3, a main storage device 4, a monitor 5, a bus 6, a memory 7, a video acquisition unit 10, and a video management unit 11. , A number recognition unit 12 and a video output unit 13.

  The digital VTR 1 is data containing route images taken by the inspection vehicle. The external interface 2 takes in video data of the digital VTR 1. The monitor 5 displays the video data of the digital VTR1.

  The image acquisition unit 10 can continuously capture the utility pole numbers arranged on the route along with the image of the route on which the train travels. The number recognizing unit 12 recognizes the number of the telephone pole number from the video acquired by the video acquiring unit 10 by pattern coding described later.

  The video management unit 11 has a list for storing the numbers recognized by the number recognition unit 12 and the accompanying information, for example, kilometres, sections, and the like. The main storage device 4 includes a video data file 8 and a utility pole number data file 9. In the video data file 8 and the telephone pole number data file 9, since the telephone pole number and the route video data obtained from the digital VTR 1 are recorded, the location of the failure occurrence location can be determined using the input device 3 (for example, keyboard, mouse, etc.). If the telephone pole number is input, the route image data corresponding to the telephone pole number can be read and reproduced on the monitor 5. Hereinafter, a process of recognizing the acquired number in the number recognition unit 12 and the like will be described.

  FIG. 2 is a diagram illustrating an example of a recognition range of utility pole numbers (numerals). The number recognition range 20 is composed of 5 × 3 (total 15) cells. The route video management system 100 sums up the “weights” set for each cell 30 when the number of the telephone pole number is taken into the number recognition range 20, and determines whether or not the total value exceeds a predetermined value. By making the determination, it is possible to pattern-code the numbers taken into the number recognition range 20 in units of cells.

  The number of cells (division unit) in the number recognition range 20 shown in FIG. 2 is a unit necessary for recognizing numbers and is not limited to this. For example, the number of cells may be configured in order to make a pattern pattern more finely. Further, the division unit may be changed according to the complexity of the recognition target. Further, the aspect ratio of the number recognition range 20 may be arbitrarily set. In consideration of “total weight” described later, the minimum division unit is preferably “vertical 3 × horizontal 3”. Note that the numbers shown on the lower and left sides of the drawing indicate the number of pixels to be described later.

  FIG. 3 is a diagram illustrating an example of pixel weighting. In FIG. 3, each of the cells shown in FIG. 2 is indicated by a bold rectangle. The cell 30 is composed of 3 × 4 pixels (12 pixels in total). The cell 30 is given a value indicating “weight” for recognizing a number.

  A large “weight” (for example, “10”) is set in the center portion (first region) of the cell 30, and a small “weight” (for example, “1”) is set in the side portion (third region) of the cell 30. Is set. Further, an intermediate “weight” (for example, “4”) between the center and the side of the cell 30 is set at the corner (second region) of the cell 30. Therefore, in the number recognition range 20 shown in FIG. 2, the “weight” is set to the same value for the pixels in the portion where each cell is adjacent. The reason why the “weight” is different for each region is to clearly recognize the number taken into the number recognition range 20 of FIG. 2, for example, the number “2” across cells. That is, in order to clarify the difference between an intended image (such as “line”) and an unintended image (such as noise), unlike mosaic processing that averages the total value of each cell. As a result, when the captured number is displayed at the center of each cell, the route image management system 100 can recognize a characteristic part of the captured number.

  The weight value is not limited to these values, and can be set arbitrarily. Further, the number of pixels of the cell 30 is not limited to this, and the cell 30 may be composed of more pixels.

  FIG. 4 is a diagram illustrating an example of a cell when a pixel pattern is acquired. As an example, the number “2” is captured in the pixel pattern 40. The numbers can be displayed in units of pixels of each cell, the portion where the number is shown (the portion shown in black in FIG. 4) indicates the lighting state, and the other portions are in the non-lighting state. Is shown.

  FIG. 5 is a diagram illustrating an example of weights corresponding to pixel patterns. For example, the number “2” shown in the number recognition range 50 is displayed in the cell 51 and the cell 53 and is not displayed in the cell 52. In the number recognition range 50, the number “2” displayed in each cell is recognized as a “weight” in pixel units.

  FIG. 6 is a diagram illustrating an example of the total cell weight. Sum the “weights” of the lit pixels in cell units. The total value of each cell indicated in the number recognition range 60 is, for example, “28”, “0”, and “7” in the cell 61, the cell 62, and the cell 63, respectively.

  FIG. 7 is a diagram illustrating an example of a cell weight threshold. The value shown in each cell in the number recognition range 70 is a value set to determine whether each cell obtained in FIG. 6 is suitable as a lighting cell. For example, the value “28” of the cell 61 shown in FIG. 6 exceeds the value “20” of the cell 71 shown in FIG. On the other hand, since the value of the cell 62 shown in FIG. 6 is “0”, it is not recognized as a lighted cell. Further, although the value of the cell 63 shown in FIG. 6 is “7”, since the value of the cell 73 shown in FIG. 7 is “20”, it is not recognized as a lighted cell. In this way, by setting a threshold value for each cell, it is possible to discriminate between characteristic portions and non-characteristic portions of the numbers displayed in the number recognition range 70. Note that the threshold value of each cell is not limited to this, and can be arbitrarily set.

  FIG. 8 is a diagram illustrating an example of cells that are colored according to cell thresholds. For each cell in the number recognition range 80, only cells that exceed the threshold value in FIG. 7 are in a lit state (the state that is shaded in FIG. 8). For example, the cell 81 is in a lighting state because it exceeds the threshold value of the cell 71, but the cell 82 and the cell 83 are in a no-light state because they do not exceed the threshold value.

  FIG. 9 is a diagram illustrating an example of a pattern-coded cell. Each cell in the number recognition range 90 shows a state in which each cell shown in FIG. 8 is turned on or off (pattern code) is replaced with bit information. That is, in FIG. 8, the lit cell is replaced with “1” indicating “with bit”, and the unlit cell is replaced with “0” indicating “without bit”. For example, the cell 91 is replaced with “1” because the cell 81 is lit, and the cells 92 and 93 are replaced with “0” because the cell 92 and the cell 93 are not lit.

  FIG. 10 is a diagram illustrating an example of a bit string. The cell information 101 indicates a state in which, for example, a number from 0 to e is assigned as a number indicating a bit string, from the lower left to the upper right in the number recognition range 20 shown in FIG. In each cell of the cell information 101, a portion surrounded by a bold line indicates a cell in which the feature of the captured telephone pole number (numeral) appears, and is used in a comparison process described later.

  The bit string 102 of the pattern code is obtained by arranging the values of the cells replaced with “1” or “0” in FIG. 9 in the order of bits from 0 to e. For example, when the captured number is “2”, the bit string “0”, “1”, “2”, “3”, “7”, “8”, “b”, “c”, “d” , And “e” are in the lighting state, the value of the bit is “111100011001111”.

  The comparison information 103 indicates only a characteristic part of the pattern code with a 7-digit code. The 7-digit code is indicated by “0”, “3”, “5”, “6”, “7”, “8”, “b”, for example, in the cell surrounded by the thick line in the cell information 101. Equivalent to. By setting such characteristic portions, redundancy can be given to the route image management system 100 and high-speed processing can be performed. That is, even when the pattern code is lost for some reason, it is possible to recognize the telephone pole number taken in only by the characteristic part. In addition, by converting only characteristic parts into a dictionary, processing can be performed at a higher speed than when pattern codes are converted into a dictionary. For example, when the fetched number is “2”, “0”, “3”, “7”, “8”, and “b” of the comparison information 103 are in a lighting state. Therefore, the route video management system 100 can recognize the number as “2” without depending on the bit string 102 of the pattern code by recognizing only the lighted cell.

  FIG. 11 is a flowchart illustrating an example of a flow for determining the number recognition position. The route image management system 100 starts binary conversion processing of the captured utility pole number (number) (step S1). The position of the utility pole number (number) displayed on the monitor 5, that is, the recognition position of the utility pole number (number) is set (step S2), and the recognition process of the utility pole number (number) is started (step S3). When the recognized utility pole number (number) is different from the previously recognized number (step S4, Yes), the recognition position of the utility pole number (number) is reset (step S5). When the recognition position of the utility pole number (number) is within the regulation (Yes in step S6), it is determined whether the utility pole number (number) is normally recognized. When it is recognized normally (step S7, Yes), it notifies that the utility pole number (number) has been recognized and ends (step S8).

  When the recognized utility pole number (number) is the same as the previously recognized utility pole number (number) (step S4, No), the route video management system 100 resets the recognition position of the utility pole number (number). End the process without When the utility pole number (number) is not recognized normally (step S7, No), the process is terminated without notifying that the utility pole number (number) is recognized. Further, when the recognition position of the utility pole number (numeral) is not specified (No in Step S6), the processes after Step S2 are repeated.

  FIG. 12 is a flowchart illustrating an example of a decision flow for binary conversion of color information. When the route image management system 100 determines that the color information of the captured utility pole number (number) is equal to or greater than a predetermined threshold (Yes in step S10), the route image management system 100 sets a value “255” indicating lighting (step S11). . When the binary conversion is completed (step S13, Yes), the process is terminated.

  When the route video management system 100 determines that the color information of the captured utility pole number (number) is less than the predetermined threshold (No in step S10), the route video management system 100 sets a value “0” indicating no light (step S12). ). If the binary conversion has not been completed (No at Step S13), the processes after Step S10 are repeated until the processing is completed.

  FIG. 13 is a flowchart illustrating an example of a determination flow for recognizing utility pole numbers (numerals). The route image management system 100 acquires the “weight” of the pixel for each cell (step S32) when the predetermined pixel is lit for the captured utility pole number (number) (step S31, Yes). The “weight” is summed for each time (step S33). If the “weight” of each cell is equal to or greater than a predetermined threshold (Yes at Step S34), the bit at the cell position of the pattern code is set to ON “1” (Step S35). When the scanning of all the pixels has been completed (step S37, Yes), the number corresponding to the pattern code is acquired and the process is terminated (step S38).

  The route image management system 100 performs the process of step S37 to scan the next pixel when a predetermined pixel is not lit for the captured utility pole number (number) (step S31, No).

  If the “weight” of the summed cells is less than the predetermined threshold (No at step S34), the bit at the cell position of the pattern code is set to OFF “0” (step S36). If the scanning of all the pixels has not been completed (No at Step S37), the processes after Step S31 are repeated until the scanning of all the pixels is completed.

  As described above, according to the route video management system 100 of the first embodiment, a cell composed of pixels set with “weight” is subjected to a “weight” total for each cell, and the captured utility pole Since the numbers (numbers) can be pattern-coded, the numbers displayed on the route image can be automatically recognized. In addition, since the utility pole number (number) can be recognized using the characteristic part of the cell, the processing can be performed at a higher speed than the pattern code is made into a dictionary. Further, since three different “weights” are set in the cell, it is possible to clearly recognize the captured number.

  As described above, the route video management system according to the present invention is useful for a video management system that automatically recognizes a number embedded as a video in a video frame as digital information.

It is a figure which shows an example of a structure of the route image | video management system concerning Embodiment 1. It is a figure which shows an example of the recognition range of a telephone pole number (number). It is a figure which shows an example of weighting of the lighting pixel in a cell. It is a figure which shows an example of the cell at the time of pixel pattern acquisition. It is a figure which shows an example of the weight corresponding to a pixel pattern. It is a figure which shows an example of the weight total of a cell. It is a figure which shows an example of the weight threshold value of a cell. It is a figure which shows an example of the cell colored by the cell threshold value. It is a figure which shows an example of the cell by which the pattern encoding was carried out. It is a figure which shows an example of a bit string. It is a flowchart which shows an example of the determination flow of a number recognition position. It is a flowchart which shows an example of the decision flow of the binary conversion of color information. It is a flowchart which shows an example of the determination flow of telephone pole number (number) recognition.

Explanation of symbols

1 Digital VTR
2 External interface 3 Input device 4 Main storage device 5 Monitor 6 Bus 7 Memory 8 Video data file 9 Telephone pole number data file 10 Video acquisition unit 11 Video management unit 12 Number recognition unit 13 Video output unit 20, 50, 60, 70, 80 , 90 Number recognition range 30, 51, 52, 53, 61, 62, 63, 71, 72, 73, 81, 82, 83, 91, 92, 93 Cell 40 Pixel pattern 100 Route video management system 101 Cell information 102 pattern Bit string of code 103 Comparison information

Claims (4)

In a route video management system that recognizes video output from video playback equipment ,
A numeral recognition range surrounding the digits, is composed of a plurality of cells, the weight of each pixel corresponding to numbers of the plurality of cells is taken to digit recognition range composed of a plurality of pixels for each of the cell It is determined for each cell whether or not the total value exceeds a predetermined threshold value, and a part of the cells within the number recognition range and each cell representing the number characteristic portion is in a lighting state A number recognizing unit that generates a bit string composed of a code indicating whether it is in a no-light state or not and recognizes a number attached to a power pole installed along a route on which the train travels;
A route image management system characterized by   The number recognition range is a cell in which a first region indicating a center portion of the cell, a second region indicating a corner portion of the cell, and a third region indicating a side portion of the cell are defined. A cell in which the weight of the pixel in the first region is set larger than the weight of the pixel in the second region, and the weight of the pixel in the second region is set larger than the weight of the pixel in the third region. The route image management system according to claim 1, further comprising:   The route image management system according to claim 1, wherein the number recognition range includes cells in which the weights of pixels adjacent to each cell are set to the same value. In a video management method for recognizing numbers embedded in video output from a video playback device,
A step of defining a number recognition range surrounding the number, wherein the number recognition range includes a plurality of cells, and the plurality of cells includes a plurality of pixels;
A summing step of summing the weight value of each pixel corresponding to the number captured in the number recognition range for each cell;
A determination step of determining, for each cell, whether or not the summed value exceeds a predetermined threshold;
A generation step of generating a bit string composed of a code indicating whether a part of the cells within the number recognition range and the cell representing the character part of the number is in a lighting state or a non-lighting state;
A video management method comprising:

Images