An automatic detection system for abnormal dirtiness of ship cleaning robots

Abstract

The present invention relates to the field of image processing technology, and in particular to an automatic detection system for dirt anomalies of a ship cleaning robot. In each frame of a ship bottom image, the system first identifies impact pixels, and marks the rest as non-impact pixels. The distance from the non-impact pixel to the impact pixel is calculated to determine the pixel of silt particles. A window is constructed for each non-impact pixel, and the proportion of silt particles in the window is calculated. A window to be adjusted and a normal window are screened out. The density of silt particles at the edge pixels of the window to be adjusted is analyzed to determine the extended edge pixels and the extended edge. According to the proportion of silt particles in the window to be adjusted, the density of silt particles at the extended edge pixels, and the extended edge, the window to be adjusted is expanded to a normal window, and a normal ship bottom image is further obtained. Finally, the normal ship bottom image is used to clean the ship bottom. The present invention improves the accuracy of identifying dirt on the ship bottom by analyzing the grayscale values ​​of all pixels at each frame of the ship bottom image.

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to an automatic detection system for abnormal dirtiness of a ship cleaning robot. Background Art

[0002] During the voyage, the bottom of the ship may be covered with dirt due to the adhesion of mud, moss, etc. The accumulation of dirt on the bottom of the ship will not only increase the resistance of the ship's navigation and increase fuel consumption, but may also change the hydrodynamic characteristics of the hull and affect the stability and safety of the ship during navigation. Therefore, it is particularly important to clean the dirt on the bottom of the ship in a timely manner.

[0003] Since ships are immersed in water most of the time, relying on divers to clean and maintain the bottom of the ship is not only time-consuming and labor-intensive, but also poses certain safety risks. In order to facilitate the identification and cleaning of ship bottom dirt, underwater robots are now used instead of manual labor to clean the ship bottom dirt, which improves the safety and efficiency of ship bottom cleaning. However, when underwater robots detect ship bottom dirt, they may be affected by suspended sand particles in the water and high-pressure water jets, resulting in unclear video footage, which in turn causes inaccurate identification of ship bottom dirt. Summary of the invention

[0004] The invention provides an automatic dirt abnormality detection system for a ship cleaning robot to solve the existing problems.

[0005] The present invention discloses a ship cleaning robot dirt abnormality automatic detection system adopts the following technical scheme:

[0006] An embodiment of the present invention provides a system for automatically detecting abnormal dirtiness of a ship cleaning robot, the system comprising the following modules:

[0007] Image acquisition module: used to obtain each frame of the ship bottom image;

[0008] Impact location analysis module: used to construct several straight line segments for each pixel point in each frame of the ship bottom image, and obtain the impact pixel point according to the number of pixels and gray value of each straight line segment of each pixel point;

[0009] Sediment particle analysis module: used to record the pixels other than the impact pixels of each frame of the ship bottom image as non-impact pixels; according to the distance from each non-impact pixel to the impact pixel of each frame of the ship bottom image, a number of sediment particle pixels are obtained;

[0010] Extended edge analysis module: used to construct a window of each non-impact pixel point in each frame of the ship bottom image; obtain the proportion of sediment particles in the window of each non-impact pixel point according to the number of sediment particle pixels in the window of each non-impact pixel point in each frame of the ship bottom image; select a number of windows to be adjusted and normal windows according to the proportion of sediment particles in the window of each non-impact pixel point in each frame of the ship bottom image; obtain the sediment density of each edge pixel point of each window to be adjusted according to the distance between all sediment particle pixels in each window to be adjusted and the distance to the edge of the window to be adjusted; obtain the extended edge pixel points and extended edges of each window to be adjusted according to the sediment density of each edge pixel point of each window to be adjusted in each frame of the ship bottom image;

[0011] Image processing and analysis module: used to expand each window to be adjusted into a normal window according to the proportion of sediment particles in each window to be adjusted in each frame of the ship bottom image, the sediment density of the extended edge pixel points of each window to be adjusted, and the extended edge; and obtain each frame of normal ship bottom image according to the normal window in each frame of the ship bottom image;

[0012] Cleaning sequence analysis module: used to clean the bottom of the ship according to each frame of normal bottom image.

[0013] Furthermore, the step of constructing a plurality of straight line segments for each pixel point in each frame of the ship bottom image and obtaining the impact pixel point according to the number of pixels and the gray value of each straight line segment of each pixel point comprises:

[0014] First The first frame of the ship bottom image The pixel point is the starting point. Draw a ray through the pixels adjacent to the pixel point, and get the Several rays of pixels;

[0015] In the The pixel On the ray, The line segment from the pixel point to the edge of the bottom image is recorded as A straight line segment with pixels;

[0016] The first The first frame of the ship bottom image The sum of the differences in the grayscale values ​​of all adjacent pixels on all straight line segments of the pixel is recorded as The first frame of the ship bottom image The pixels are the probability of impacting the pixels;

[0017] The first The first frame of the ship bottom image The pixel is the product of the probability of the impact pixel and the gray value, recorded as The first frame of the ship bottom image Pixels are the final possibilities of impacting pixels;

[0018] Each pixel point of each frame of the ship bottom image is the pixel point with the maximum final probability of being the impact pixel point, and is recorded as the impact pixel point of each frame of the ship bottom image.

[0019] Furthermore, the obtaining of a plurality of sediment particle pixels according to the distance from each non-impact pixel to the impact pixel of each frame of the ship bottom image comprises:

[0020] According to the distance from each non-impact pixel point to the impact pixel point in each frame of the ship bottom image, the possibility that each non-impact pixel point in each frame of the ship bottom image is a sediment particle pixel point is obtained;

[0021] like is greater than the preset first threshold, then The first frame of the ship bottom image non-impact pixel point, denoted as The pixel points of sediment particles in the frame bottom image; For the The first frame of the ship bottom image The probability that a non-impact pixel is a sediment particle pixel.

[0022] Furthermore, the possibility that each non-impact pixel point in each frame of the ship bottom image is a sediment particle pixel point is obtained according to the distance from each non-impact pixel point to the impact pixel point in each frame of the ship bottom image, including:

[0023] In the In the frame of the ship bottom image, The standard deviation of the grayscale values ​​of the non-impact pixel and all its adjacent non-impact pixels is The ratio of the Euclidean distance between the non-impact pixel and the impact pixel is recorded as The first frame of the ship bottom image The probability that a non-impact pixel is a sediment particle pixel.

[0024] Furthermore, the method of obtaining the proportion of sediment particles in the window of each non-impact pixel point according to the number of sediment particle pixels in the window of each non-impact pixel point of each frame of the ship bottom image comprises:

[0025] In the In the frame of the ship bottom image, The ratio of the number of sediment particle pixels in the window with non-impact pixels to the number of non-impact pixels in the window is recorded as The first frame of the ship bottom image The proportion of sediment particles in the window of non-impact pixels.

[0026] Furthermore, the method of screening out a plurality of windows to be adjusted and normal windows according to the proportion of sediment particles in the window of each non-impact pixel point of each frame of the ship bottom image includes:

[0027] Jordi The first frame of the ship bottom image If the proportion of sediment particles in the window of the non-impact pixel is greater than the preset second threshold, the The first frame of the ship bottom image The window of non-impact pixels is denoted as The window to be adjusted for the frame bottom image; if The first frame of the ship bottom image If the proportion of sediment particles in the window of the non-impact pixel point is less than or equal to the preset second threshold, the first The first frame of the ship bottom image The window of non-impact pixels is denoted as Normal window for frame bottom image.

[0028] Furthermore, the method of obtaining the sediment density of each edge pixel point of each window to be adjusted according to the distance between all sediment particle pixels of each window to be adjusted and the distance to the edge of the window to be adjusted in each frame of the ship bottom image comprises:

[0029] In the The first frame of the ship bottom image In the window to be adjusted, The pixel of a sediment particle and its nearest predecessor The sum of the reciprocals of the distances to the pixels of the sediment particles is recorded as The first frame of the ship bottom image The window to be adjusted The density of the distribution of sediment particle pixels; is a preset quantity threshold;

[0030] In the In the frame of the ship bottom image, calculate the The window to be adjusted The density of the distribution of the first sediment particle pixel is similar to that of the The window to be adjusted The edge pixels and The ratio of the distance between the pixels of the sediment particles is The density of the distribution of all the sediment particle pixels in the window to be adjusted is the same as that in the first The window to be adjusted The sum of the ratios of the distances between the edge pixels and all the sediment particle pixels is recorded as The first frame of the ship bottom image The window to be adjusted The sediment density of each edge pixel.

[0031] Further, the step of obtaining the extended edge pixel points and extended edges of each window to be adjusted according to the sediment density of each edge pixel point of each window to be adjusted in each frame of the ship bottom image comprises:

[0032] The first The first frame of the ship bottom image The edge pixel point with the smallest sediment density in the window to be adjusted is recorded as The first frame of the ship bottom image The extended edge pixels of the window to be adjusted;

[0033] The first The first frame of the ship bottom image The edge where the extended edge pixel of the window to be adjusted is located is recorded as the extended edge.

[0034] Further, the step of expanding each window to be adjusted into a normal window according to the proportion of sediment particles in each window to be adjusted in each frame of the ship bottom image, the sediment density of the extended edge pixel points of each window to be adjusted, and the extended edge includes:

[0035] In the In the frame of the ship bottom image, The proportion of sediment particles in the window to be adjusted is The normalized value of the ratio of the sediment density of the extended edge pixel points of the window to be adjusted is recorded as The first frame of the ship bottom image The extent of the window to be adjusted;

[0036] The first The first frame of the ship bottom image The sum of the product of the expansion degree of the window to be adjusted and the preset window length threshold and the preset window length threshold is recorded as The first frame of the ship bottom image The extended length of the window to be adjusted;

[0037] According to the extended length of each window to be adjusted of each frame of the ship bottom image, the extended edge of each window to be adjusted of each frame of the ship bottom image is extended outward to obtain a correction window of each window to be adjusted;

[0038] According to the method of obtaining the proportion of sediment particles in the window of each non-impact pixel point, the proportion of sediment particles in each correction window is obtained;

[0039] In the In the frame of the ship bottom image, if If the proportion of sediment particles in the first correction window is less than or equal to the preset second threshold, A correction window, denoted as The normal window of the frame bottom image; if the The proportion of sediment particles in the first correction window is greater than the preset second threshold value, and the first correction window is obtained according to the acquisition method of the correction window of each window to be adjusted. The next correction window is the correction window of the correction window, until the When any of the correction windows is corrected again to a normal window, the first The normal window corresponding to the correction window.

[0040] Furthermore, the cleaning of the ship bottom according to each frame of normal ship bottom image includes:

[0041] According to the gray value of the pixel points in each frame of the normal ship bottom image, the canny operator is used to perform edge detection on each frame of the normal ship bottom image to obtain a number of edge pixel points of each frame of the normal ship bottom image;

[0042] According to all edge pixel points of each frame of normal ship bottom image, the contour of each frame of normal ship bottom image is extracted using the contour extraction algorithm, and several closed contour areas of each frame of normal ship bottom image are recorded as muddy areas;

[0043] The reciprocal of the distance from the pixel point at the center of each muddy area of ​​each frame of the normal ship bottom image to the impact pixel point of each frame of the normal ship bottom image is recorded as the cleaning priority coefficient of the muddy area;

[0044] The underwater robot cleans the ship bottom according to the cleaning priority coefficients of all muddy areas in each frame of normal ship bottom image from high to low.

[0045] The beneficial effects of the technical solution of the present invention are:

[0046] Get each frame of the ship bottom image; construct several straight line segments for each pixel in each frame of the ship bottom image, obtain the impact pixel according to the number of pixels and grayscale value of each straight line segment of each pixel, and accurately identify the impact pixel. Record the pixels other than the impact pixel of each frame of the ship bottom image as non-impact pixels; obtain several sediment particle pixels according to the distance from each non-impact pixel to the impact pixel of each frame of the ship bottom image, accurately determine the sediment particle pixels, and provide key data for subsequent processing. In each frame of the ship bottom image, construct a window for each non-impact pixel; according to the number of sediment particle pixels in the window of each non-impact pixel of each frame of the ship bottom image, obtain the sediment particle ratio of each non-impact pixel window, accurately reflecting the sediment particle ratio in each window. According to the sediment particle ratio of each non-impact pixel window of each frame of the ship bottom image, select several windows to be adjusted and normal windows, and accurately select the windows that need to be adjusted. According to the distance between all the mud and sand particle pixels of each window to be adjusted in each frame of the bottom image and the distance to the edge of the window to be adjusted, the mud and sand density of each edge pixel of each window to be adjusted is obtained, which accurately reflects the mud and sand density of each edge pixel and provides key data for obtaining the extended edge later. According to the mud and sand density of each edge pixel of each window to be adjusted in each frame of the bottom image, the extended edge pixel and extended edge of each window to be adjusted are obtained, which improves the effect of subsequent expansion. According to the proportion of mud and sand particles in each window to be adjusted in each frame of the bottom image, the mud and sand density of the extended edge pixel of each window to be adjusted and the extended edge, each window to be adjusted is expanded into a normal window, ensuring the efficiency and accuracy of image processing. According to the normal window in each frame of the bottom image, each frame of the normal bottom image is obtained; according to each frame of the normal bottom image, the bottom is cleaned. The present invention improves the accuracy of bottom dirt recognition by analyzing the grayscale values ​​of all pixels of each frame of the bottom image. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 This is a module flow chart of an automatic detection system for abnormal dirtiness of a ship cleaning robot according to the present invention;

[0049] Figure 2 This is a flow chart of identifying and cleaning dirt on the bottom of a ship in this embodiment;

[0050] Figure 3 This is a schematic diagram of the outward expansion of the window to be adjusted provided in this embodiment. DETAILED DESCRIPTION

[0051] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation method, structure, features and effects of the automatic detection system for abnormal dirtiness of a ship cleaning robot proposed by the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0052] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0053] The specific scheme of the automatic detection system for abnormal dirtiness of a ship cleaning robot provided by the present invention is described in detail below with reference to the accompanying drawings.

[0054] See also Figure 1 , which shows a module flow chart of a system for automatically detecting abnormal dirtiness of a ship cleaning robot provided by an embodiment of the present invention, the system comprising the following modules:

[0055] Image acquisition module 101:

[0056] This module is used to obtain each frame of the ship bottom image.

[0057] The underwater robot continuously shoots video during the ship bottom detection process to obtain a ship bottom video; and captures a ship bottom image of each frame in the ship bottom video;

[0058] It should be noted that underwater robots are usually composed of multiple parts such as cameras, sensors, and cleaning systems.

[0059] Impact location analysis module 102:

[0060] This module is used to construct several straight line segments for each pixel in each frame of the ship bottom image, and obtain the impact pixel point according to the number of pixels and grayscale value of each straight line segment of each pixel point.

[0061] It should be noted that: when high-pressure water impacts the bottom of the ship, water mist is formed. The concentration of water mist is the highest at the impact position, and the concentration of water mist in the surrounding area gradually decreases. Therefore, when identifying the pixels of sediment particles suspended in the water, the closer the sediment particle pixels are to the impact position of the high-pressure water, the higher the possibility that they are sediment particles. When identifying the pixels of sediment particles far away from the impact position, some non-sediment particle pixels may be misjudged as sediment particles due to insufficient light and other reasons, that is, the possibility that the identified sediment particle pixels are sediment particles is low. Therefore, it is necessary to first analyze the impact position of high-pressure water on the bottom of the ship.

[0062] It is further necessary to explain that when high-pressure water impacts the bottom of the ship, the water mist formed shows a characteristic in which the grayscale value gradually decreases from the impact point to the surrounding areas on the bottom image, and the grayscale value of the pixel at the impact point should be greater than the grayscale value of the pixel at other positions.

[0063] It is further necessary to explain that due to the interference of other factors such as the reflection of sediment particles in the water, pixels with large gray values ​​may also be generated, so further screening is required based on the change in gray values. Extending from the pixel point of the impact point to the surrounding area, the gray value should show a trend of gradually decreasing as a whole, but the pixels of sediment particles are scattered and do not meet this feature, so according to the gray changes of each pixel in 8 directions, the possibility of it being an impact pixel in terms of gray change is calculated.

[0064] First The first frame of the ship bottom image For example, taking the first pixel The pixel point is taken as the starting point, and a ray is drawn through its adjacent pixel points to obtain the A number of rays from each pixel.

[0065] Through the above process, a number of rays of each pixel point of each frame of the ship bottom image are obtained.

[0066] In the The pixel On the ray, The line segment from the pixel point to the edge of the bottom image is recorded as A straight line segment with pixels.

[0067] Through the above process, a number of straight line segments of each pixel point of each frame of the ship bottom image are obtained.

[0068] The first The first frame of the ship bottom image The specific calculation formula corresponding to the probability that a pixel point is an impact pixel point is:

[0069]

[0070] in, Indicates The first frame of the ship bottom image The pixels are the probability of impacting the pixels; Indicates The first frame of the ship bottom image The number of straight line segments per pixel; Indicates The first frame of the ship bottom image The pixel The number of pixels of a straight line segment; Indicates The first frame of the ship bottom image The pixel The first The gray value of each pixel; No. The first frame of the ship bottom image The pixel The first The gray value of a pixel.

[0071] What needs to be explained is: and Proportional, The larger the value of The first frame of the ship bottom image The more pixels there are, the more likely they are impact pixels. In the calculation process, the The traversal starts from the pixel point.

[0072] Through the above process, we obtain The first frame of the ship bottom image The pixels are the probability of impacting the pixels.

[0073] Each pixel point of each frame of the ship bottom image is operated according to the above process to obtain the possibility that each pixel point of each frame of the ship bottom image is an impact pixel point.

[0074] The first The first frame of the ship bottom image The pixel is the product of the probability of the impact pixel and the gray value, recorded as The first frame of the ship bottom image Pixels are the ultimate possibility of impacting pixels.

[0075] Through the above process, it is obtained that each pixel point of each frame of the ship bottom image is the final possibility of the impact pixel point.

[0076] Each pixel point of each frame of the ship bottom image is the pixel point with the maximum final probability of being the impact pixel point, and is recorded as the impact pixel point of each frame of the ship bottom image.

[0077] It should be noted that: when there are multiple pixels with the highest final probability that each pixel point of each frame of the ship bottom image is an impact pixel point, one of the pixels is randomly selected as the impact pixel point.

[0078] Sediment particle analysis module 103:

[0079] This module is used to record the pixels other than the impact pixels of each frame of the ship bottom image as non-impact pixels; and obtain a number of sediment particle pixels according to the distance from each non-impact pixel to the impact pixel of each frame of the ship bottom image.

[0080] It should be noted that the grayscale of the suspended sediment particles in the ship bottom image is usually quite different from that of the pixels on the ship bottom, so its grayscale performance is similar to that of noise pixels. Therefore, the larger the standard deviation of the grayscale value of the pixel in the 8-neighborhood of the pixel, the more likely it is that the pixel is a sediment particle pixel.

[0081] It should be further explained that the closer the pixel is to the impact point, the more likely it is to be a real sediment particle pixel. Therefore, the distance between each pixel and the impact point pixel is combined to calculate the possibility that each pixel is a suspended sediment particle pixel.

[0082] All pixels other than the impact pixels in each frame of the ship bottom image are recorded as non-impact pixels.

[0083] First The first frame of the ship bottom image For example, the non-impact pixel The first frame of the ship bottom image The specific calculation formula for the probability that a non-impact pixel is a sediment particle pixel is:

[0084]

[0085] in, Indicates The first frame of the ship bottom image The probability that a non-impact pixel is a sediment particle pixel, Indicates The first frame of the ship bottom image The standard deviation of the grayscale values ​​of a non-impact pixel and all its adjacent non-impact pixels is: Indicates The first frame of the ship bottom image The Euclidean distance between the non-impact pixel and the impact pixel.

[0086] What needs to be explained is: and Inversely proportional, The larger the value of The first frame of the ship bottom image The smaller the probability that a non-impact pixel is a sediment particle pixel; and Proportional, The larger the value of The first frame of the ship bottom image The more likely a non-impact pixel is a sediment particle pixel.

[0087] Through the above process, we obtain The first frame of the ship bottom image The probability that a non-impact pixel is a sediment particle pixel.

[0088] Each non-impact pixel point of each frame of the ship bottom image is calculated according to the above process to obtain the possibility that each non-impact pixel point of each frame of the ship bottom image is a pixel point of a sediment particle.

[0089] like is greater than the preset first threshold, then The first frame of the ship bottom image non-impact pixel point, denoted as The pixel points of the sediment particles in the frame bottom image. In this embodiment, the first threshold value preset is 0.6, which is used as an example for description.

[0090] Through the above process, a number of sediment particle pixels in each frame of the ship bottom image are obtained.

[0091] Extended edge analysis module 104:

[0092] This module is used to construct a window for each non-impact pixel in each frame of the ship bottom image; obtain the proportion of sediment particles in the window of each non-impact pixel according to the number of sediment particle pixels in the window of each non-impact pixel in each frame of the ship bottom image; screen out several windows to be adjusted and normal windows according to the proportion of sediment particles in the window of each non-impact pixel in each frame of the ship bottom image; obtain the sediment density of each edge pixel of each window to be adjusted according to the distance between all sediment particle pixels in each window to be adjusted and the distance to the edge of the window to be adjusted; obtain the extended edge pixels and extended edges of each window to be adjusted according to the sediment density of each edge pixel of each window to be adjusted in each frame of the ship bottom image.

[0093] It should be noted that the water mist formed when the robot cleans the dirt on the bottom of the ship will cause the image to be blurred, so it is impossible to accurately identify the dirt on the bottom of the ship in this state. Therefore, it is necessary to perform defogging on each frame of the bottom image.

[0094] It is further necessary to explain that when using the dark channel prior algorithm to defog the bottom image, the dark channel value of the pixel needs to be calculated based on the minimum value of the three channels in the window corresponding to each pixel. The color of the sediment particles and the bottom of the ship is quite different, and the degree of influence of the water mist formed by high-pressure water is also inconsistent. Therefore, the proportion of suspended sediment particles in the processing window will have a greater impact on the calculation result of the dark channel value. Therefore, the window size needs to be adjusted to ensure that the proportion of sediment particle pixels in the window is as small as possible.

[0095] In each frame of the ship bottom image, take each non-impact pixel as the center and set The initial window size is used to construct the window of each non-impact pixel. is the preset window length threshold. The preset window length threshold in this embodiment is The value is 15, and this example is used for description.

[0096] Still The first frame of the ship bottom image For example, the non-impact pixel The ratio of the number of sediment particle pixels in the window with non-impact pixels to the number of non-impact pixels in the window is recorded as The first frame of the ship bottom image The proportion of sediment particles in the window of non-impact pixels.

[0097] Through the above process, the proportion of sediment particles in the window of each non-impact pixel point of each frame of the ship bottom image is obtained.

[0098] Jordi The first frame of the ship bottom image If the proportion of sediment particles in the window of the non-impact pixel is greater than the preset second threshold, the The first frame of the ship bottom image The window of non-impact pixels is denoted as The window to be adjusted for the frame bottom image. The first frame of the ship bottom image If the proportion of sediment particles in the window of the non-impact pixel point is less than or equal to the preset second threshold, the first The first frame of the ship bottom image The window of non-impact pixels is denoted as The normal window of the frame bottom image. In this embodiment, the second threshold value preset is 0.3, which is used as an example for description.

[0099] Through the above process, several windows to be adjusted are obtained for each frame of the ship bottom image.

[0100] It should be noted that the suspended sand particles in the water are caused by the high-pressure water being flushed to the places with sand, which disperses the originally settled sand. Therefore, in the places where there is already sediment, after being flushed with high-pressure water, there will be more sand particle pixels in the bottom image, while the originally clean places will be less affected. Therefore, the expansion edge and expansion degree of the window to be adjusted can be determined according to the distribution of sand particle pixels in the window to be adjusted.

[0101] First The first frame of the ship bottom image The window to be adjusted For example, the pixel point of the sediment particle The pixel of a sediment particle and its nearest predecessor The sum of the reciprocals of the distances to the pixels of the sediment particles is recorded as The first frame of the ship bottom image The window to be adjusted The density of the distribution of sediment particle pixels. is a preset quantity threshold. The preset quantity threshold in this embodiment is The value is 5, and this is used as an example for description.

[0102] Through the above process, the density of the distribution of each pixel point of the sediment particles in each window to be adjusted in each frame of the ship bottom image is obtained.

[0103] It should be noted that: the closer the sediment particle pixel points with high distribution density are to the edge of the window to be adjusted, the denser the sediment particle pixel points near the edge of the window to be adjusted; therefore, the sediment density of each edge pixel point of the window to be adjusted is calculated based on the distribution density of each sediment particle pixel point and the distance from the edge of the window to be adjusted.

[0104] The first The first frame of the ship bottom image The window to be adjusted The specific calculation formula corresponding to the sediment density of each edge pixel is:

[0105]

[0106] in, Indicates The first frame of the ship bottom image The window to be adjusted The density of sediment at each edge pixel; Indicates The first frame of the ship bottom image The number of pixels of sediment particles in the window to be adjusted; Indicates The first frame of the ship bottom image The window to be adjusted The density of the distribution of sediment particle pixels; Indicates The first frame of the ship bottom image The window to be adjusted The edge pixels and The distance between the pixels of the sediment particles.

[0107] What needs to be explained is: and Proportional, The larger the value of The first frame of the ship bottom image The window to be adjusted The density of sediment at each edge pixel is greater.

[0108] Through the above process, we obtain The first frame of the ship bottom image The window to be adjusted The sediment density of each edge pixel.

[0109] Each edge pixel point of each window to be adjusted in each frame of the ship bottom image is calculated according to the above process to obtain the sediment density of each edge pixel point of each window to be adjusted in each frame of the ship bottom image.

[0110] It should be noted that in order to reduce the proportion of sediment particle pixels in the expanded window to be adjusted, it is necessary to expand outside the edge where the edge pixel with the smallest sediment density in the window to be adjusted is located. Therefore, the expansion edge of the window to be adjusted is determined according to the sediment density of all edge pixels of each window to be adjusted.

[0111] The first The first frame of the ship bottom image The edge pixel point with the smallest sediment density in the window to be adjusted is recorded as The first frame of the ship bottom image The extended edge pixel points of the window to be adjusted. If there are multiple edge pixels with the smallest sediment density, any one is selected for analysis.

[0112] Through the above process, the extended edge pixel points of each window to be adjusted in each frame of the ship bottom image are obtained.

[0113] The first The first frame of the ship bottom image The edge where the extended edge pixel of the window to be adjusted is located is recorded as the extended edge.

[0114] Image processing and analysis module 105:

[0115] This module is used to expand each window to be adjusted into a normal window according to the proportion of sediment particles in each window to be adjusted in each frame of the bottom image, the sediment density of the extended edge pixels of each window to be adjusted, and the extended edge; and obtain each frame of normal bottom image according to the normal window in each frame of the bottom image.

[0116] It should be noted that: according to the above process, the larger the proportion of sediment particles in the window to be adjusted, and the smaller the sediment density of the extended edge pixels of the window to be adjusted, the greater the degree of expansion required.

[0117] First The first frame of the ship bottom image For example, the window to be adjusted The first frame of the ship bottom image The specific calculation formula corresponding to the expansion degree of the window to be adjusted is:

[0118]

[0119] in, Indicates The first frame of the ship bottom image The extent of the window to be adjusted; Indicates The first frame of the ship bottom image The proportion of sediment particles in the window to be adjusted; Indicates The first frame of the ship bottom image The density of sediment at the extended edge pixels of the window to be adjusted; represents the linear normalization function.

[0120] Through the above process, we obtain The first frame of the ship bottom image The extent of the window to be adjusted.

[0121] Each window to be adjusted in each frame of the ship bottom image is operated according to the above process to obtain the expansion degree of each window to be adjusted in each frame of the ship bottom image.

[0122] The first The first frame of the ship bottom image The sum of the product of the expansion degree of the window to be adjusted and the preset window length threshold and the preset window length threshold is recorded as The first frame of the ship bottom image The extended length of the window to be adjusted.

[0123] Through the above process, the extended length of each window to be adjusted in each frame of the ship bottom image is obtained.

[0124] According to the extended length of each window to be adjusted of each frame of the bottom image, the extended edge of each window to be adjusted of each frame of the bottom image is extended outward to obtain the correction window of each window to be adjusted. Figure 3 This is a schematic diagram of the outward expansion of the window to be adjusted provided in this embodiment. Figure 3 In the figure, A represents the extended edge, C represents the extended length, and B represents the extended edge after extension.

[0125] According to the method of obtaining the proportion of sediment particles in the window of each non-impact pixel point, the proportion of sediment particles in each correction window is obtained.

[0126] First The first frame of the ship bottom image Take the correction window as an example. If the proportion of sediment particles in the first correction window is less than or equal to the preset second threshold, A correction window, denoted as The normal window of the frame bottom image; if the The proportion of sediment particles in the first correction window is greater than the preset second threshold value, and the first correction window is obtained according to the acquisition method of the correction window of each window to be adjusted. The next correction window is the correction window of the correction window, until the When any of the correction windows is corrected again to a normal window, the first The normal window corresponding to the correction window.

[0127] Each corrected window of each frame of the ship bottom image is operated according to the above process to obtain a plurality of normal windows of each frame of the ship bottom image.

[0128] A dark channel a priori algorithm is used to operate each normal window of each frame of the ship bottom image to obtain several frames of normal ship bottom images. The dark channel a priori algorithm is a well-known technology, and the specific method is not introduced here.

[0129] Cleaning sequence analysis module 106:

[0130] This module is used to clean the bottom of the ship according to each frame of normal bottom image.

[0131] It should be noted that: in the process of cleaning, the underwater robot should reduce repeated cleaning of the bottom of the ship as much as possible and improve the working efficiency of the underwater robot. It is necessary to reasonably plan the working path of the underwater robot according to the position of the muddy area in each frame of the normal bottom image. Find the pixel point at the center of each muddy area, and calculate the cleaning priority of each muddy area according to the distance between the pixel point at the center of the muddy area and the current impact pixel point. Among them, the flowchart of the identification and cleaning of the bottom of the ship is as follows: Figure 2 shown.

[0132] According to the grayscale value of the pixel points in each frame of the normal ship bottom image, the canny operator is used to perform edge detection on each frame of the normal ship bottom image to obtain a number of edge pixel points of each frame of the normal ship bottom image. The canny operator is a well-known technology and the specific method is not introduced here.

[0133] According to all edge pixel points of each frame of normal ship bottom image, contour extraction algorithm is used to extract contour of each frame of normal ship bottom image, and several closed contour areas of each frame of normal ship bottom image are recorded as muddy areas. The contour extraction algorithm is a well-known technology, and the specific method is not introduced here.

[0134] The reciprocal of the distance from the pixel point at the center of each muddy area of ​​each frame of the normal ship bottom image to the impact pixel point of each frame of the normal ship bottom image is recorded as the cleaning priority coefficient of the muddy area.

[0135] The underwater robot cleans the ship bottom according to the cleaning priority coefficients of all muddy areas in each frame of normal ship bottom image from high to low.

[0136] It should be noted that: in this embodiment, when the denominator in the formula is 0, the denominator is set to 1 to ensure that the formula is valid, and this example is used for description.

[0137] So far, the present invention is completed.

[0138] In summary, in an embodiment of the present invention, in each frame of the ship bottom image, the impact pixel points are first identified, and the rest are marked as non-impact pixel points. The distance from the non-impact pixel point to the impact pixel point is calculated, and the pixel point of the silt particle is determined. A window is constructed for each non-impact pixel point, and the proportion of silt particles in the window is calculated. The window to be adjusted and the normal window are screened out. The silt density of the edge pixel points of the window to be adjusted is analyzed, and the extended edge pixel points and the extended edge are determined. According to the proportion of silt particles in the window to be adjusted, the silt density of the extended edge pixel points, and the extended edge, the window to be adjusted is expanded to a normal window, and a normal ship bottom image is further obtained. Finally, the normal ship bottom image is used to clean the ship bottom. The present invention improves the accuracy of ship bottom dirt identification by analyzing the grayscale values ​​of all pixels of each frame of the ship bottom image.

[0139] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An automatic detection system for abnormal dirtiness of a ship cleaning robot, characterized in that: The system includes the following modules: Image acquisition module: used to obtain each frame of the ship bottom image; Impact location analysis module: used to construct several straight line segments for each pixel point in each frame of the ship bottom image, and obtain the impact pixel point according to the number of pixels and gray value of each straight line segment of each pixel point; Sediment particle analysis module: used to record the pixels other than the impact pixels of each frame of the ship bottom image as non-impact pixels; according to the distance from each non-impact pixel to the impact pixel of each frame of the ship bottom image, a number of sediment particle pixels are obtained; Extended edge analysis module: used to construct a window of each non-impact pixel point in each frame of the ship bottom image; obtain the proportion of sediment particles in the window of each non-impact pixel point according to the number of sediment particle pixels in the window of each non-impact pixel point in each frame of the ship bottom image; and select a number of windows to be adjusted and normal windows according to the proportion of sediment particles in the window of each non-impact pixel point in each frame of the ship bottom image; According to the distances between all the sediment particle pixels of each window to be adjusted in each frame of the ship bottom image and the distance to the edge of the window to be adjusted, the sediment density of each edge pixel of each window to be adjusted is obtained; According to the sediment density of each edge pixel point of each window to be adjusted in each frame of the ship bottom image, the extended edge pixel point and the extended edge of each window to be adjusted are obtained; Image processing and analysis module: used to expand each window to be adjusted into a normal window according to the proportion of sediment particles in each window to be adjusted in each frame of the ship bottom image, the sediment density of the extended edge pixel points of each window to be adjusted, and the extended edge; and obtain each frame of normal ship bottom image according to the normal window in each frame of the ship bottom image; Cleaning sequence analysis module: used to clean the bottom of the ship according to each frame of normal bottom image.

2. According to claim 1, a ship cleaning robot dirt abnormality automatic detection system is characterized in that: The step of constructing a plurality of straight line segments for each pixel point in each frame of the ship bottom image and obtaining the impact pixel point according to the number of pixels and the gray value of each straight line segment of each pixel point comprises: First The first frame of the ship bottom image The pixel point is the starting point. Draw a ray through the pixels adjacent to the pixel point, and get the Several rays of pixels; In the The pixel On the ray, The line segment from the pixel point to the edge of the bottom image is recorded as A straight line segment with pixels; The first The first frame of the ship bottom image The sum of the grayscale value differences of all adjacent pixels on all straight line segments of the pixel is recorded as The first frame of the ship bottom image The pixels are the probability of impacting the pixels; The first The first frame of the ship bottom image The probability that the first pixel is the impact pixel is The first frame of the ship bottom image The product of the grayscale values ​​of pixels is recorded as The first frame of the ship bottom image Pixels are the final possibilities of impacting pixels; Each pixel point of each frame of the ship bottom image is the pixel point with the maximum final probability of being the impact pixel point, and is recorded as the impact pixel point of each frame of the ship bottom image.

3. According to claim 1, the ship cleaning robot dirt abnormality automatic detection system is characterized by: The method of obtaining a plurality of sediment particle pixels according to the distance from each non-impact pixel to the impact pixel of each frame of the ship bottom image comprises: According to the distance from each non-impact pixel point to the impact pixel point in each frame of the ship bottom image, the possibility that each non-impact pixel point in each frame of the ship bottom image is a sediment particle pixel point is obtained; like is greater than the preset first threshold, then The first frame of the ship bottom image non-impact pixel point, denoted as The pixel points of sediment particles in the frame bottom image; For the The first frame of the ship bottom image The probability that a non-impact pixel is a sediment particle pixel.

4. According to claim 3, the ship cleaning robot dirt abnormality automatic detection system is characterized in that: The possibility that each non-impact pixel point in each frame of the ship bottom image is a sediment particle pixel point is obtained according to the distance from each non-impact pixel point to the impact pixel point in each frame of the ship bottom image, including: In the In the frame of the ship bottom image, The standard deviation of the grayscale values ​​of the non-impact pixel and all its adjacent non-impact pixels is The ratio of the Euclidean distance between the non-impact pixel and the impact pixel is recorded as The first frame of the ship bottom image The probability that a non-impact pixel is a sediment particle pixel.

5. According to claim 1, the ship cleaning robot dirt abnormality automatic detection system is characterized by: The method of obtaining the percentage of sediment particles in the window of each non-impact pixel point according to the number of sediment particles in the window of each non-impact pixel point of each frame of the ship bottom image comprises: In the In the frame of the ship bottom image, The ratio of the number of sediment particle pixels in the window with non-impact pixels to the number of non-impact pixels in the window is recorded as The first frame of the ship bottom image The proportion of sediment particles in the window of non-impact pixels.

6. According to claim 1, the ship cleaning robot dirt abnormality automatic detection system is characterized by: The method of selecting a plurality of windows to be adjusted and normal windows according to the proportion of sediment particles in the window of each non-impact pixel point of each frame of the ship bottom image includes: Jordi The first frame of the ship bottom image If the proportion of sediment particles in the window of the non-impact pixel is greater than the preset second threshold, the The first frame of the ship bottom image The window of non-impact pixels is denoted as The window to be adjusted for the frame bottom image; if The first frame of the ship bottom image If the proportion of sediment particles in the window of the non-impact pixel point is less than or equal to the preset second threshold, the first The first frame of the ship bottom image The window of non-impact pixels is denoted as Normal window for frame bottom image.

7. According to claim 1, the ship cleaning robot dirt abnormality automatic detection system is characterized by: The method of obtaining the sediment density of each edge pixel point of each window to be adjusted according to the distance between all sediment particle pixels of each window to be adjusted and the distance to the edge of the window to be adjusted in each frame of the ship bottom image comprises: In the The first frame of the ship bottom image In the window to be adjusted, The pixel of a sediment particle and its nearest predecessor The sum of the reciprocals of the distances to the pixels of the sediment particles is recorded as The first frame of the ship bottom image The window to be adjusted The density of the distribution of sediment particle pixels; is a preset quantity threshold; In the In the frame of the ship bottom image, calculate the The window to be adjusted The density of the distribution of the first sediment particle pixel is similar to that of the The window to be adjusted The edge pixels and The ratio of the distance between the pixels of the sediment particles is The density of the distribution of all the sediment particle pixels in the window to be adjusted is the same as that in the first The window to be adjusted The sum of the ratios of the distances between the edge pixels and all the sediment particle pixels is recorded as The first frame of the ship bottom image The window to be adjusted The sediment density of each edge pixel.

8. According to claim 1, the ship cleaning robot dirt abnormality automatic detection system is characterized by: The method of obtaining the extended edge pixel points and extended edges of each window to be adjusted according to the sediment density of each edge pixel point of each window to be adjusted in each frame of the ship bottom image comprises: The first The first frame of the ship bottom image The edge pixel point with the smallest sediment density in the window to be adjusted is recorded as The first frame of the ship bottom image The extended edge pixels of the window to be adjusted; The first The first frame of the ship bottom image The edge where the extended edge pixel of the window to be adjusted is located is recorded as the extended edge.

9. According to claim 1, the ship cleaning robot dirt abnormality automatic detection system is characterized by: The step of expanding each window to be adjusted into a normal window according to the proportion of sediment particles in each window to be adjusted in each frame of the ship bottom image, the sediment density of the extended edge pixel points of each window to be adjusted, and the extended edge comprises: In the In the frame of the ship bottom image, The proportion of sediment particles in the window to be adjusted is The normalized value of the ratio of the sediment density of the extended edge pixel points of the window to be adjusted is recorded as The first frame of the ship bottom image The extent of the window to be adjusted; The first The first frame of the ship bottom image The sum of the product of the expansion degree of the window to be adjusted and the preset window length threshold and the preset window length threshold is recorded as The first frame of the ship bottom image The extended length of the window to be adjusted; According to the extended length of each window to be adjusted of each frame of the ship bottom image, the extended edge of each window to be adjusted of each frame of the ship bottom image is extended outward to obtain a correction window of each window to be adjusted; According to the method of obtaining the proportion of sediment particles in the window of each non-impact pixel point, the proportion of sediment particles in each correction window is obtained; In the In the frame of the ship bottom image, if If the proportion of sediment particles in the first correction window is less than or equal to the preset second threshold, A correction window, denoted as The normal window of the frame bottom image; if the The proportion of sediment particles in the first correction window is greater than the preset second threshold value, and the first correction window is obtained according to the acquisition method of the correction window of each window to be adjusted. The next correction window is the correction window of the correction window, until the When any of the correction windows is corrected again to a normal window, the first The normal window corresponding to the correction window.

10. The ship cleaning robot dirt abnormality automatic detection system according to claim 1, characterized in that: The cleaning of the ship bottom according to each frame of normal ship bottom image comprises: According to the gray value of the pixel points in each frame of the normal ship bottom image, the canny operator is used to perform edge detection on each frame of the normal ship bottom image to obtain a number of edge pixel points of each frame of the normal ship bottom image; According to all edge pixel points of each frame of normal ship bottom image, the contour of each frame of normal ship bottom image is extracted using the contour extraction algorithm, and several closed contour areas of each frame of normal ship bottom image are recorded as muddy areas; The reciprocal of the distance from the pixel point at the center of each muddy area of ​​each frame of the normal ship bottom image to the impact pixel point of each frame of the normal ship bottom image is recorded as the cleaning priority coefficient of the muddy area; The underwater robot cleans the ship bottom according to the cleaning priority coefficients of all muddy areas in each frame of normal ship bottom image from high to low.

Images