JP2012185061A - Water level detection device, water level detection system, and water level detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that sometimes a true draft line cannot be detected even when a projection pattern of luminance is used in the case that a false draft line has darkness similar to the true draft line and in the case that the vicinity of the false draft line is extremely bright.SOLUTION: Projection addition is performed to an edge image generated from a photographed image including a draft line, and linear components are extracted on the basis of a pattern of the projection addition. Difference images to be a difference between the photographed image and each of respective past images prescribed frames before the photographed image are generated, a set of pixels having the luminance exceeding a prescribed threshold among cumulative difference images generated by cumulatively adding the luminance that respective pixels of the respective difference images have is determined as a water surface area, and the linear component not overlapping with the water surface area among the extracted linear components is determined as the true draft line. Thus, detection accuracy of the draft line is improved.

Description

  The present invention relates to a water level detection device, a water level detection system, and a water level detection method for analyzing image data captured by a camera and detecting the water level.

  In a conventional water level detection device, a water level is detected by identifying a water line where a predetermined object such as a bridge girder and a water surface intersect from image data (input image) taken by a camera that monitors a river or the like. . The draft line performs projection addition processing (one-dimensional projection processing) on edge image data generated by extracting a sharp edge of luminance change of each pixel data constituting the input image, and the projection addition pattern (edge (Accumulated result).

  However, in such image processing, in addition to the actual water line (true water line), false water lines caused by waves on the water surface, shadows of bridge girders, and the like may be extracted. For this reason, it may be difficult to identify the true water line from the cumulative edge result. In the conventional water level detection device, paying attention to the influence of the shadow inserted into the water surface, in addition to the cumulative result of the edge, the projection pattern of the luminance of the input image (cumulative result of the luminance image) is created simultaneously, By analyzing the depth of the V-shaped drop in the middle, false detection due to a false water line generated on the water surface was eliminated (for example, see Patent Document 1).

JP 2007-212238 A

  In the conventional water level detection device, when the false water line has a darkness similar to the true water line, or when the vicinity of the false water line is extremely bright, the V-shaped drop obtained from the cumulative result of the luminance image From this depth, it was difficult to detect fake waterlines. For this reason, there is a problem that accurate water level detection is not always possible.

  This invention was made in order to solve the above problems, and it aims at providing the water level detection apparatus, the water level detection system, and the water level detection method which improved the detection accuracy of the waterline.

  The water level detection device according to the present invention, an image input means for inputting a photographed image including a water line, generates an edge image from the photographed image, and performs projection addition in a specific direction in a region where the edge image is set, A linear component extraction image processing unit that extracts a linear component based on the projection addition pattern, and generates a difference image that is a difference between the captured image and each past image of the captured image before a predetermined frame. Difference accumulation means for accumulating the luminance of each pixel of the image to generate an accumulated difference image, and water surface judgment for determining a set of pixels having luminance exceeding a predetermined threshold as the water surface area among the accumulated difference images Means, and a water line determining means for determining, as a true water line, a straight line component that does not overlap the water surface area among the extracted straight line components.

  The water level detection device, the water level detection system, and the water level detection method of the present invention can improve the detection accuracy of the water line.

It is a block diagram which shows the main structure means of the water level detection apparatus of Embodiment 1 of this invention. It is a block diagram (a) which shows the detail of the waterline extraction image processing means 3 of Embodiment 1 of this invention, a figure which shows the intensity distribution (d) of a picked-up image (b), an edge image (c), and a projection addition value. . It is a figure which shows the image (c) explaining the difference result of the present image (a), background image (b), and these images about Embodiment 1 of this invention. It is an image explaining the difference accumulation result which concerns on Embodiment 1 of this invention. The image (a) which shows the water surface determined using the change area | region binarization method about Embodiment 1 of this invention, and the image (b) which shows the water surface determined using the change area | region binarization method line expansion method FIG. It is an image which shows several waterlines and water surface area | region about Embodiment 1 of this invention. (A) shows the case where one of the two water lines is in contact with the water surface area, and (b) shows the case where both of the two water lines are in contact with the water surface area. It is a block diagram which shows the main structure means of the water level detection apparatus of Embodiment 2 of this invention.

Embodiment 1 FIG.
Hereinafter, Embodiment 1 of the present invention will be described in detail with reference to the drawings. FIG. 1 is a configuration diagram showing main components of the water level detection device according to the first embodiment of the present invention. FIG. 2 is a block diagram (a) showing details of the waterline extraction image processing means 3 according to the first embodiment of the present invention, a photographed image (b), an edge image (c), and an intensity distribution (d) of the projected cumulative value. FIG. FIG. 3 is a diagram showing a current image (a), a background image (b), and an image (c) for explaining a difference result between these images according to the first embodiment of the present invention. FIG. 4 is an image for explaining the difference accumulation result according to the first embodiment of the present invention. FIG. 5 is an image (a) showing the water surface determined using the change region binarization method and an image showing the water surface determined using the change region binarization method line expansion method according to Embodiment 1 of the present invention. It is a figure which shows (b). FIG. 6 is an image showing a plurality of waterlines and water surface areas according to the first embodiment of the present invention. (A) shows the case where one of the two water lines is in contact with the water surface area, and (b) shows the case where both of the two water lines are in contact with the water surface area.

  As shown in FIG. 1, the water level detection apparatus includes an image information input I / F 1, an image information input unit 2, a waterline extraction image processing unit 3, a storage unit 4, a current image storage unit 5, and a background image storage unit. Means 6, difference cutout means 7, difference accumulation means 8, water surface determination means 9, waterline determination means 10, result output means 11, and image information output I / F 12.

  The image information input I / F 1 receives video information (hereinafter referred to as a video image signal) captured by a camera or the like that monitors a river or the like.

  The image information input means 2 converts the video image signal input from the image information input I / F 1 into digital data. For example, the luminance of the pixels constituting the image is converted into 8-bit multi-value data (monochrome image data) to obtain multi-value image data (current image data). The current image data output from the image information input unit 2 is input to the waterline extraction image processing unit 3, the current image storage unit 5, and the background image storage unit 6, and the process branches.

  The waterline (straight line component) extraction image processing means 3 performs a waterline extraction process on the latest current image data input from the image information input means 2 and extracts one or a plurality of straight line components (draft lines). In the case of extracting a plurality of straight line components, a legitimate straight line component is a true water line, and a straight line component caused by a wave or a shadow of a bridge girder is a false water line. Detailed description of the waterline extraction process will be described later.

  The storage unit 4 stores water level information, image information, and the like obtained by converting the position of the waterline into a water level within a preset search range that is a range for edge extraction.

  The current image storage unit 5 stores the latest current image data input from the image information input unit 2.

  The background image storage unit 6 stores the current image data input from the image information input unit 2 as background image data for comparison with the current image data based on a predetermined rule.

  As for the predetermined rule, for example, the past image data input from the image information input unit 2 4 frames before (about 130 ms before) the current image data input from the image information input unit 2 and the accumulated past image data is used as the background image. The rules are as follows. Hereinafter, the past image data four frames before is described as a background image, but the present invention is not limited to this.

  The difference cutout unit 7 calculates a luminance difference for each pixel between the current image data input from the current image storage unit 5 and each background image data input from the background image storage unit 6. (Difference image) is generated. The difference is generated, for example, when a luminance difference is generated between the current image data and the background image data due to the movement of waves generated on the water surface. Detailed description of the difference image will be described later.

  The difference accumulating unit 8 accumulates each difference data generated by the difference cutout unit 7 and generates an accumulated difference image. Accumulation means a process of simply adding each obtained difference image for each pixel for a predetermined time (for example, 1 second). Detailed description of the cumulative difference image will be described later.

  The water surface determination unit 9 determines (predicts) an area that is a water surface based on the accumulated difference image output from the difference accumulation unit 8. Detailed description of the water surface determination will be described later.

  The draft line determination unit 10 includes a contact determination unit 101 and a highest level determination unit 102, and the draft line extracted by the draft line extraction image processing unit 3 is valid based on the water surface area determined by the water surface determination unit 9. Determine whether there is. A detailed description of the determination method will be described later.

  The result output means 11 outputs the water level information of the true water line determined by the water line determination means 10. For example, a superimposition image in which the determination result information is superimposed on the image as characters is generated and output to the monitor device, or for the personal computer, the determination result information is incorporated into a LAN (Local Area Network) packet. It can also be delivered.

  The image information output I / F 12 outputs the result output unit 11.

  The waterline extraction process in the waterline extraction image processing means 3 will be described with reference to FIG. Here, in order to simplify the description, a case where there is no fake water line will be described. As shown in FIG. 2A, the waterline extraction image processing unit 3 includes an edge extraction unit 31, a range designation unit 32, a straight line detection unit 33, and a water level detection unit 34. The edge extraction unit 31 performs edge extraction processing on the image data as shown in FIG. 2B input from the image information input unit 2 to generate an edge image as shown in FIG. The range designating unit 32 designates a search range, which is a range in which edge extraction is performed, from the edge information of the edge image output from the edge extracting unit 31 and extracts edge information within the range. Note that the edge extraction means 31 may perform edge extraction processing only within the search range of the input image.

  The straight line detection means 33 performs one-dimensional projection processing in a specific direction on the edge information within the range specified by the range specification means 32, and the peak of the intensity distribution (pattern) of the projection addition value as shown in FIG. A linear component is detected based on the value. When a plurality of straight line components are extracted, a false water line exists. The water level detection means 34 determines that the straight line component detected by the straight line detection means 33 is a draft line, and outputs the position information of the draft line and the absolute value information of the water level corresponding to the position of the draft line stored in the storage means 4.

  A difference image generated by the difference cutout unit 7 will be described with reference to FIG. (A) is the current image, (b) is the background image four frames before the current image, and (c) is the difference image generated by the difference calculation between the current image and the background image within the search range. For (c), the difference image is assumed to be superimposed on the current image for convenience of explanation. The difference method in the example of FIG. 3 is a 4-frame difference method, and the obtained difference image is a 4-frame difference image. In addition, since the difference calculation is performed only within the search range here, the pixel indicating the position of the difference is generated only within the search range. It doesn't matter.

  The accumulated difference image generated by the difference accumulation unit 8 will be described with reference to FIG. The image in FIG. 4 is obtained by superimposing a cumulative difference image on the current image for convenience of explanation. In one difference image before accumulation as shown in FIG. 3C, which is generated by the difference cutout means 7, the difference that is generated is created by a motion such as a wave at a certain moment between the current image and the background image. It is a small difference, and the position where the difference occurs moves and changes, so it is not stable. However, by accumulating each difference image for a predetermined time, as shown in FIG. 4, it is possible to aggregate the differences that occur sporadically and irregularly into one image. As a result, even if the difference image of each image is very small, that is, even if the luminance difference between the current image and the background image is small, the difference image between the two images can be changed by accumulating the difference images for a predetermined time. It can be remarkably raised.

  The water surface determination in the water surface determination means 9 is demonstrated using FIG. The water surface determination is performed by, for example, a change area binarization method or a change area binarization method line expansion method.

  First, the change area binarization method will be described. In the change area binarization method, binarization processing is performed with a predetermined threshold on the accumulated difference image having multiple values in each pixel. By the binarization processing, a pixel having a value larger than a predetermined threshold (a pixel having a large difference) is determined as “1”, and a pixel having a value smaller than the predetermined threshold (a pixel having a small difference) is determined as “0”. . Here, a set of pixels that are “1” is defined as a change region. The change region binarization method assumes that this change region is the “water surface”. As shown in FIG. 5A, when the change area binarization method is used, the hatched portion that is the change area is determined to be the water surface.

  Next, the change region binarization method line expansion method will be described. In contrast to the change area binarization method in which the change area is determined to be “water surface”, the change area binarization method line enlargement method is an area in which pixels having a value of “1” after binarization are enlarged in the line direction. Is different in that it is determined as “water surface”. Therefore, in the change area binarization method line expansion method, a line having even one pixel having a value of “1” after binarization is assumed to be the water surface in units of the line. In general, a range larger than the change area binarization method is determined as the “water surface”. As shown in FIG. 5B, when the change region binarization method line enlargement method is used, the broken line region is determined as the water surface.

  A method for determining a water line in the water line determining means 10 will be described with reference to FIG. The contact determination unit 101 determines whether or not the plurality of waterlines extracted by the waterline extraction image processing unit 3 are in contact with (overlap with) the water surface area determined by the water surface determination unit 9. In FIG. 6A, the water line 51 is not in contact with the water surface area, and the water line 52 is in contact with the water surface area. The contact determination means 101 determines that the waterline 52 that contacts the water surface area is a false waterline and rejects it, determines that the waterline 51 does not contact the water surface area is a true waterline, and determines the position of the true waterline. The water level information converted into the water level is output to the result output means 11. In addition, the water line 52 is a false water line that is erroneously recognized because the waves are aligned on a straight line.

  In FIG.6 (b), both the waterline 53 and the waterline 54 are contacting the water surface area | region. In this case, the contact determination unit 101 determines that both the water lines are fake water lines and rejects them, and outputs an error signal indicating the conclusion that no water line is found to the highest level determination means 102.

  The uppermost determination means 102 to which the error signal has been input determines that the water line that is present at the uppermost position among the plurality of water lines that are in contact with the water surface region is the true water line, and rejects the other water lines. That is, of the draft line 53 and the draft line 54, the highest draft line 53 is determined as a true draft line, and the draft line 54 is determined as a false draft line. Then, the water line 54 is rejected, and the water level information obtained by converting the position of the true water line 53 into the water level is output to the result output unit 11. In the example of FIG. 6, the case where two draft lines are extracted has been described. For example, even if there are three or more draft lines, it is possible to reject a false draft line and determine a true draft line.

  As described above, according to the first embodiment, the water surface area is determined based on the difference image data obtained from the current image data and the background image data (past image data), and the water surface area is touched among the plurality of water lines. It is possible to improve the detection accuracy of the true draft line by rejecting the draft line as a false draft line.

  As an example of the rule for generating the background image data, the example in which the past image data of the past four frames is used as the background image data has been described. The past image data includes, for example, the past one from the time when the current image data is output. It also includes image data obtained by averaging the past image data accumulated during the second.

  The water level detection unit 34 may be provided at the subsequent stage of the water line determination unit 10, and may output water level information obtained by converting the position of the water line determined by the water line determination unit 10 into a water level to the result output unit 11.

Embodiment 2. FIG.
Hereinafter, a second embodiment of the present invention will be described in detail with reference to the drawings. FIG. 7 is a block diagram showing main components of the water level detection device according to the second embodiment of the present invention. Parts corresponding to those of the first embodiment are denoted by the same reference numerals as those in FIG. The second embodiment is different from the first embodiment in that an additional mask unit 13 is added before the draft line image processing unit 3. Hereinafter, a flow of a series of image processing operations from the start (image information input I / F1) to the end (image information output I / F12) as described in the first embodiment to the determination of the waterline will be described. This is called one processing cycle.

  When a false draft line is found in the draft determination means 10 in a treatment cycle (assumed to be N-1 treatment cycle) one cycle before a certain treatment cycle (assuming N treatment cycle), the treatment cycle (N-1 cycle) ) Prevents misrecognition of the waterline, and there is no problem. However, as described above, the false waterline is extracted because there is a linear component caused by waves, shadows, etc., and this linear component continues to exist at that position unless there is a particularly large water level change. It is common. Therefore, in many cases, a fake draft line is continuously extracted for a certain period of time in the vicinity of a fake draft line once extracted.

  For these reasons, even in the processing cycle after the N-1 processing cycle, the draft extraction image processing means 3 continuously performs the extraction work without distinguishing between the false draft line and the true draft line. The load is proportional to the number of draft lines to be extracted. There is always one true waterline, but there is no upper limit on the number of false waterlines. Therefore, when the number of false drafts increases due to river conditions, the work in the draft extraction image processing means 3 becomes a heavy load, the time required for one processing cycle increases, and the real-time nature of the draft recognition work is impaired. End up.

  The additional mask means 13 acquires the position information of the false water line determined by the water line determination means 10 in the N-1 processing cycle in the N processing cycle, and adds the position information to the position information during the predetermined processing cycle from the N processing cycle. Based on this, the current image data input from the image input means 2 is subjected to mask processing at a position near the false waterline.

  The neighborhood here may be, for example, about 8 lines (pixel row) above and below the position where the false draft line determined in the N-1 processing cycle exists, but is not particularly limited. Further, the number of processing cycles corresponding to, for example, 30 seconds is appropriate as the predetermined processing cycle, but there is no particular limitation. When the predetermined processing cycle elapses, the additional mask means 13 removes the mask. In this way, image processing is initialized at regular intervals, and the correct mask is updated again in the next processing cycle.

  The waterline extraction image processing means 3 performs a waterline extraction process on the current image data subjected to the mask processing by the additional mask means 13. In the vicinity of the fake water line that is considered to have some linear component in the N-1 processing cycle, there is no image because the mask processing is performed. Therefore, in the N processing cycle, the waterline extraction image processing means 3 does not perform the waterline extraction processing for the portion where the mask processing is performed.

  As described above, according to the second embodiment, the additional mask means 13 performs mask processing in the vicinity of the position based on the position information of the false waterline in the processing cycle one cycle before. There is no need to repeatedly extract fake waterlines, the work load is reduced, and real-time performance of waterline recognition work can be secured without increasing the time required for one processing cycle.

1 Image information input I / F
2 Image information input means 3 Waterline extraction image processing means 31 Edge extraction means 32 Range designation means 33 Straight line detection means 34 Water level detection means 4 Storage means 5 Current image storage means 6 Background image storage means 7 Difference extraction means 8 Difference accumulation means 9 Water surface Determination means 10 Drunk line determination means 101 Contact determination means 102 Top level determination means 11 Result output means 12 Image information output I / F
13 Additional mask means

Claims (6)

An image input means for inputting a photographed image including a water line;
An edge image is generated from the photographed image, a projection addition is performed in a specific direction in a set region of the edge image, and a linear component extraction image processing unit that extracts a linear component based on the projection addition pattern;
Difference accumulation means for generating a difference image that is a difference between the photographed image and each past image of the photographed image before a predetermined frame, and cumulatively adding the luminance of each pixel of each difference image to generate a cumulative difference image When,
A water surface determination means for determining a set of pixels having luminance exceeding a predetermined threshold among the cumulative difference images as a water surface region;
A water level detection device comprising: a water line determining means for determining, as a true water line, a linear component that does not overlap the water surface region among the extracted linear components.   The water level detection device according to claim 1, wherein the water surface determination unit determines a set of horizontal pixel rows of pixels having luminance exceeding the predetermined threshold as a water surface region. The water line determining means is a contact determining means for outputting an error signal when all the extracted linear components are included in the water surface area,
3. The water level according to claim 1, further comprising: a highest level determination unit that determines, as a water line, a linear component that is uppermost among the extracted linear components when the error signal is input. Detection device.   An additional mask means for masking a predetermined time in the vicinity of the position of the straight line component is inputted from the draft line determining means, and the position information of straight line components other than the straight line component corresponding to the true water line among the extracted straight line components is provided. The water level detection device according to any one of claims 1 to 3. A photographing device for photographing an area including a water line;
A water level detection system comprising: the water level detection device according to any one of claims 1 to 3, wherein the water level detection device performs processing for taking an image photographed by the photographing device and determining a water line. A linear component extraction step for generating an edge image from a photographed image including a water line, performing projection addition in a specific direction in a set region of the edge image, and extracting a linear component based on the projection addition pattern;
A difference calculation step for calculating a difference between the captured image and each past image before a predetermined frame of the captured image;
A difference accumulation step of calculating by adding the luminance of each pixel of each difference calculation result,
As a result of the cumulative difference calculation, a water surface determination step of determining a set of pixels having luminance exceeding a predetermined threshold as a water surface region;
A water level detection method comprising: determining a straight line component that is not included in the water surface region among the extracted straight line components as a draft line.

Images