An underwater image enhancement method for underwater robot cleaning

Abstract

The present invention relates to the field of image processing technology, and in particular to an underwater image enhancement method for underwater robot cleaning, comprising: obtaining a plurality of continuous color images of the bottom surface of a ship; obtaining all suspected contaminated areas in the color images of the bottom surface of the ship by clustering the plurality of color images of the bottom surface of the ship; screening all suspected contaminated areas in the color images of the bottom surface of the ship according to color complexity and texture complexity to obtain all contaminated areas; obtaining the degree of enhancement to be obtained for the contaminated area according to the color complexity and the distribution of the contaminated area and the surrounding contaminated areas; enhancing the color image of the bottom surface of the ship based on the degree of enhancement to obtain the enhanced image of the bottom surface of the ship. The present invention improves the enhancement effect of the underwater image of the underwater robot cleaning.

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to an underwater image enhancement method for underwater robot cleaning. Background Art

[0002] Bottom cleaning is one of the most critical steps in ship maintenance, because when large ships are sailing in the ocean, their bottoms are immersed in water for a long time, and the seawater will continuously wash the bottom of the ship with various substances, resulting in the accumulation of a lot of barnacles, seaweed, shells and other substances on the bottom of the ship; these substances will not only affect the ship's sailing speed and fuel consumption, but some biological pollutants will secrete corrosive substances, accelerate the aging of the bottom of the ship, and lead to an increase in the ship's docking maintenance cycle, an increase in the cost of cleaning the bottom of the ship, a significant shortening of the ship's effective sailing time, and a reduction in its service life.

[0003] Traditional methods of cleaning the bottom of a ship mainly rely on manual cleaning. In recent years, with the continuous development of robot technology, the use of underwater robots for cleaning the bottom of a ship has become a major trend. Due to underwater visibility and floating objects in the water, it is difficult for underwater robots to identify pollutants on the bottom of the ship. Therefore, it is necessary to enhance the underwater images cleaned by underwater robots to quickly obtain the polluted area on the bottom of the ship. Traditional image enhancement algorithms are based on grayscale images, and in the complex color environment underwater, it is difficult for grayscale images to accurately analyze and distinguish the polluted area on the bottom of the ship, resulting in poor enhancement effect. Summary of the invention

[0004] In order to solve the above problems, the present invention provides an underwater image enhancement method for underwater robot cleaning, the method comprising:

[0005] Acquire multiple continuous color images of the bottom surface of the ship;

[0006] By clustering multiple color images of the bottom surface of the ship, all suspected pollution areas in the color images of the bottom surface of the ship are obtained;

[0007] Performing color quantization on the color image of the bottom surface of the ship to obtain a main color image of the color image of the bottom surface of the ship; the main color image includes several main colors; obtaining the color complexity of each suspected polluted area according to the number of main colors in each suspected polluted area and the spatial distribution of the main colors; obtaining the texture complexity of the suspected polluted area according to the similarity between the suspected polluted area and the main colors in the surrounding suspected polluted areas; screening all suspected polluted areas in the color image of the bottom surface of the ship according to the color complexity and the texture complexity to obtain all polluted areas; obtaining the degree of enhancement to be obtained in the polluted area according to the color complexity and the distribution of the polluted area and the surrounding polluted areas;

[0008] The color image of the bottom surface of the ship is enhanced based on the degree to be enhanced to obtain an enhanced image of the bottom surface of the ship.

[0009] Preferably, the method of clustering a plurality of color images of the bottom surface of a ship to obtain all suspected pollution areas in the color images of the bottom surface of the ship includes the following specific methods:

[0010] For any color image of the bottom surface of a ship, the number of cluster centers and the initial cluster center positions of the color image of the bottom surface of the ship are obtained by using the elbow method; the number of cluster centers and the initial cluster center positions of the color image of the bottom surface of the ship are input into the K-means clustering algorithm to cluster the color image of the bottom surface of the ship to obtain a number of cluster clusters;

[0011] For any cluster, in any ship bottom surface color image, the Euclidean distance between the cluster center of the cluster and the center position of the ship bottom surface color image is recorded as the first distance of the cluster in the ship bottom surface color image; if the first distances of the cluster in all ship bottom surface color images are the same, the area to which the cluster belongs is regarded as a suspected pollution area.

[0012] Preferably, the color complexity of each suspected polluted area is obtained according to the number of main colors in each suspected polluted area and the spatial distribution of the main colors, including the specific method of:

[0013] For any suspected pollution area in the color image of the bottom surface of the ship, the spatial average distance and spatial distribution value of each main color are obtained according to the spatial distribution of each main color in the suspected pollution area in the main color image;

[0014] The sum of the normalized value of the mean of the spatial distribution values ​​of all main colors in the suspected pollution area and 1 is recorded as the first sum value;

[0015] The product of the number of all main colors in the suspected pollution area, the cumulative sum of the spatial average distances of all main colors, and the first sum is taken as the color complexity of the suspected pollution area.

[0016] Preferably, the obtaining of the spatial average distance and spatial distribution value of each primary color includes the following specific methods:

[0017] For any main color in the suspected pollution area, the average of the Euclidean distances between all the pixels of the main color and the centroid of the suspected pollution area is taken as the spatial average distance of the main color;

[0018] The product of the square value of the spatial average distance of the main color and π is recorded as the first product; the ratio of the first product to the area of ​​the suspected pollution area is taken as the spatial distribution value of the main color.

[0019] Preferably, the method of obtaining the texture complexity of the suspected polluted area according to the similarity between the main colors of the suspected polluted area and the surrounding suspected polluted areas includes:

[0020] For any suspected pollution area in the color image of the bottom surface of the ship, the similarity factor of the suspected pollution area is obtained by comparing the similarity of each main color in the suspected pollution area;

[0021] Any other suspected contaminated area in the color image of the bottom surface of the ship is recorded as a reference suspected contaminated area; the Euclidean distance between the cluster center of the reference suspected contaminated area and the cluster center of the suspected contaminated area is recorded as the target distance of the reference suspected contaminated area; all the reference suspected contaminated areas in the color image of the bottom surface of the ship are sorted from large to small according to the target distance to obtain a reference area sequence of the suspected contaminated area; the first 8 reference suspected contaminated areas in the reference area sequence of the suspected contaminated area are all used as target suspected contaminated areas of the suspected contaminated area; the mean of the similarity factors of all target suspected contaminated areas of the suspected contaminated area is used as the texture complexity of the suspected contaminated area.

[0022] Preferably, the method of obtaining the similarity factor of the suspected contaminated area includes:

[0023] Combining any two main colors in the suspected contaminated area as a main color combination of the suspected contaminated area;

[0024] For any main color combination in the suspected contaminated area, the degree of overlap of the main color combination is obtained according to the spatial average distance of the main colors in the main color combination;

[0025] The ratio of the overlap of the main color combination to the color distance between the two main colors in the main color combination is taken as the similarity of the main color combination; the average of the similarities of all main color combinations in the suspected contaminated area is taken as the similarity factor of the suspected contaminated area.

[0026] Preferably, the obtaining of the overlap of the main color combination includes the following specific methods:

[0027] The Euclidean distance between the spatial centers of the two main colors in the main color combination on the suspected pollution area is recorded as the spatial distance of the main color combination; the sum of the spatial average distances of the two main colors in the main color combination is recorded as the distance sum value; the ratio of the spatial distance of the main color combination to the distance sum value is recorded as the distance ratio; the difference between 1 and the distance ratio is taken as the overlap degree of the main color combination.

[0028] Preferably, the method of screening all suspected polluted areas in the color image of the bottom surface of the ship according to the color complexity and the texture complexity to obtain all polluted areas includes:

[0029] A threshold parameter T is preset, and for any suspected pollution area in the color image of the bottom surface of the ship, possible factors of the suspected pollution area are obtained according to the color complexity and texture complexity;

[0030] If the possible factor of the suspected polluted area is greater than the threshold parameter T, the suspected polluted area is recorded as a polluted area.

[0031] Preferably, the possible factors of the suspected contaminated area are obtained by:

[0032] The normalized value of the ratio of the color complexity of the suspected polluted area to the texture complexity of the suspected polluted area is used as the possible factor of the suspected polluted area.

[0033] Preferably, the degree of enhancement of the polluted area is obtained by the color complexity and the distribution of the polluted area and the surrounding polluted areas, and the specific method includes:

[0034] For any polluted area in the color image of the bottom surface of the ship, any other polluted area in the color image of the bottom surface of the ship is recorded as a reference polluted area; the Euclidean distance between the cluster center of the reference polluted area and the cluster center of the polluted area is recorded as the target reference distance of the reference polluted area; all reference polluted areas in the color image of the bottom surface of the ship are sorted from large to small according to the target reference distance to obtain a reference polluted area sequence of the polluted area; the first 8 reference polluted areas in the reference polluted area sequence of the polluted area are all used as the target polluted areas of the polluted area;

[0035] The variance of the Euclidean distance between the centroid of the polluted area and the centroid of all target polluted areas is recorded as the first variance; the cumulative sum of the Euclidean distances between the centroid of the polluted area and the centroid of all target polluted areas is recorded as the first cumulative sum; the inverse proportional value of the product of the first variance and the first cumulative sum is recorded as the first inverse proportional value; the product of the color complexity of the polluted area and the first inverse proportional value is taken as the degree to be enhanced of the polluted area.

[0036] The beneficial effects of the technical solution of the present invention are as follows: the present invention clusters multiple color images of the bottom surface of the ship to obtain all suspected polluted areas in the color image of the bottom surface of the ship; screens all suspected polluted areas in the color image of the bottom surface of the ship according to color complexity and texture complexity to obtain all polluted areas; obtains the degree of enhancement to be achieved in the polluted area through the color complexity and the distribution of the polluted area and the surrounding polluted areas; enhances the color image of the bottom surface of the ship based on the degree of enhancement to obtain the enhanced bottom surface image; thereby accurately obtaining the polluted area, enhancing different polluted areas to different degrees, and thereby improving the enhancement effect of the underwater image cleaned by the underwater robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] 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.

[0038] Figure 1 A flowchart of the steps of an underwater image enhancement method for underwater robot cleaning according to the present invention;

[0039] Figure 2 The present invention is a characteristic relationship flow chart of an underwater image enhancement method for underwater robot cleaning. DETAILED DESCRIPTION

[0040] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the underwater image enhancement method for underwater robot cleaning proposed by the present invention, its specific implementation, structure, features and effects, 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.

[0041] 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.

[0042] The specific scheme of the underwater image enhancement method for underwater robot cleaning provided by the present invention is described in detail below with reference to the accompanying drawings.

[0043] See also Figure 1 , which shows a flowchart of a method for underwater image enhancement for underwater robot cleaning provided by an embodiment of the present invention, the method comprising the following steps:

[0044] Step S001: Acquire a plurality of continuous color images of the bottom surface of the ship.

[0045] It should be noted that due to the changeable water environment, abnormal environment is easy to cause interference to the bottom pollution area in the captured image; therefore, there is a high probability that there will be small suspended matter in the underwater environment. When the underwater robot and the bottom of the ship remain relatively still, the suspended matter in the water will move in the continuous images taken in a short period of time, but the position of the real bottom pollution organisms in the image will not change.

[0046] Specifically, we first need to collect multiple continuous color images of the bottom surface of the ship. The specific process is as follows:

[0047] Before the underwater robot performs the cleaning operation, it is first fixed on the bottom surface of the ship and remains stationary for 2 seconds. During the stationary process, the underwater camera configured by the underwater robot is used to continuously shoot the bottom surface of the ship in front of the robot to obtain 4 consecutive color images of the bottom surface of the ship. The shooting cycle is 0.5 seconds, that is, 4 consecutive color images of the bottom surface of the ship are obtained.

[0048] So far, a plurality of continuous color images of the bottom surface of the ship are obtained through the above method.

[0049] Step S002: clustering multiple color images of the bottom surface of the ship to obtain all suspected pollution areas in the color images of the bottom surface of the ship.

[0050] It should be noted that since the pollutants on the bottom of the ship are divided into different types such as seaweed, shellfish, barnacles, etc., but they are all characterized by being tightly clustered; therefore, the use of clustering algorithms can well identify these marine polluting organisms; because there is a high probability that there will be smaller suspended matter in the underwater environment, when the underwater robot remains relatively still with the bottom of the ship, the suspended matter in the water will move in the continuous images taken in a short period of time, while the position of the real bottom polluting organisms in the image will not change, which will lead to small noise areas in the clustering results.

[0051] Preferably, in some implementations of the embodiments of the present invention, a specific method for obtaining all suspected pollution areas in the color images of the bottom surface of a ship by clustering multiple color images of the bottom surface of the ship is:

[0052] For any color image of the bottom surface of a ship, the number of cluster centers and the initial cluster center positions of the color image of the bottom surface of the ship are obtained by using the elbow method; the number of cluster centers and the initial cluster center positions of the color image of the bottom surface of the ship are input into the K-means clustering algorithm to cluster the color image of the bottom surface of the ship to obtain a number of cluster clusters; similarly, a number of cluster clusters of all the color images of the bottom surface of the ship are obtained;

[0053] For any cluster, in any ship bottom surface color image, the Euclidean distance between the cluster center of the cluster and the center position of the ship bottom surface color image is recorded as the first distance of the cluster in the ship bottom surface color image; if the first distances of the cluster in all ship bottom surface color images are the same, the area to which the cluster belongs is regarded as a suspected contaminated area; if the first distances of the cluster in all ship bottom surface color images are different, the area to which the cluster belongs is regarded as a noise area.

[0054] It should be noted that the elbow method and the K-means clustering algorithm are existing algorithms, and will not be described in detail in this embodiment.

[0055] So far, all suspected pollution areas in each color image of the bottom surface of the ship are obtained through the above method.

[0056] Step S003: perform color quantization on the color image of the bottom surface of the ship to obtain a main color image of the color image of the bottom surface of the ship; the main color image includes several main colors; obtain the color complexity of each suspected polluted area according to the number of main colors in each suspected polluted area and the spatial distribution of the main colors; obtain the texture complexity of the suspected polluted area according to the similarity between the suspected polluted area and the main colors in the surrounding suspected polluted areas; screen all suspected polluted areas in the color image of the bottom surface of the ship according to the color complexity and texture complexity to obtain all polluted areas; obtain the degree of enhancement to be obtained in the polluted area through the color complexity and the distribution of the polluted area and the surrounding polluted areas.

[0057] 1. Obtain the color complexity of each suspected contaminated area.

[0058] It should be noted that a normal ship's bottom is often painted, and because the bottom surface of the hull structure should be streamlined and smooth, it should show consistent color in the image; while the bottom of the ship to which marine pollution organisms attach will show relatively complex colors in the image.

[0059] Preferably, in some implementations of the embodiments of the present invention, the color image of the bottom surface of the ship is color quantized to obtain the main color image of the bottom surface of the ship; according to the number of main colors in each suspected pollution area and the spatial distribution of the main colors, the specific method for obtaining the color complexity of each suspected pollution area is:

[0060] The color image of the bottom surface of the ship is color quantized by a 64-color color quantization method to obtain a primary color image of the bottom surface color image; wherein the primary color image has several primary colors; the 64-color color quantization method is a prior art and will not be described in detail in this embodiment.

[0061] For any suspected pollution area in the color image of the bottom surface of the ship, the spatial average distance and spatial distribution value of each main color are obtained according to the spatial distribution of each main color in the suspected pollution area in the main color image;

[0062] The sum of the normalized value of the mean of the spatial distribution values ​​of all main colors in the suspected pollution area and 1 is recorded as the first sum value;

[0063] The product of the number of all main colors in the suspected pollution area, the cumulative sum of the spatial average distances of all main colors, and the first sum is taken as the color complexity of the suspected pollution area.

[0064] The specific formula is:

[0065] P=C×R×(1+norm(DS))

[0066] Where P represents the color complexity of the suspected polluted area; C represents the number of all main colors in the suspected polluted area; R represents the cumulative sum of the spatial average distances of all main colors in the suspected polluted area; DS represents the mean of the spatial distribution values ​​of all main colors in the suspected polluted area; norm() represents the linear normalization function.

[0067] It should be noted that the more primary colors contained in the suspected polluted area, the more colors the suspected polluted area is composed of, and the higher its color complexity; the cumulative sum of the spatial average distances of all primary colors in the suspected polluted area represents the discreteness of the primary colors in the suspected polluted area. The larger the value, the more chaotic the overall distribution of the primary colors in the suspected polluted area, and the higher its color complexity; the spatial distribution value of each primary color can be simply regarded as a circle formed by the center of the primary color in space and the spatial average distance. The ratio of the circle area to the image area can measure the size of the spatial distribution of a primary color. The average value of the spatial distribution of all primary colors can reflect the shape characteristics of an area. That is, the larger the average value of the spatial distribution of all primary colors, the wider the distribution range of all primary colors in the suspected polluted area. For a color-chaotic area, the larger its area, the more complex its color is.

[0068] Preferably, in some implementations of the embodiments of the present invention, the method for obtaining the spatial average distance and spatial distribution value of each primary color includes:

[0069] For any main color in the suspected pollution area, the average of the Euclidean distances between all the pixels of the main color and the centroid of the suspected pollution area is taken as the spatial average distance of the main color;

[0070] The product of the square value of the spatial average distance of the main color and π is recorded as the first product; the ratio of the first product to the area of ​​the suspected pollution area is taken as the spatial distribution value of the main color.

[0071] At this point, the color complexity of each suspected pollution area is obtained.

[0072] 2. Obtain the texture complexity of each suspected contaminated area.

[0073] It should be noted that texture can be regarded as a combination of one or more bases. The base is used to describe local grayscale information. Its combination can be random or obey a certain rule. The stronger the regularity of the basis combination, the simpler the texture characteristics. This regularity is reflected in the spatial distribution of color: the stronger the regularity, the more similar the color distribution in different areas, and the more periodic repetition occurs. The marine polluting organisms on the bottom of the ship are diverse and their distribution is often disorderly. Therefore, the texture complexity is judged by comparing the similarity of the main colors in the suspected polluted area. The lower the similarity, the higher the complexity, and the more likely it is a polluted area.

[0074] Preferably, in some implementations of the embodiments of the present invention, the specific method for obtaining the texture complexity of the suspected polluted area according to the similarity between the suspected polluted area and the main colors in the surrounding suspected polluted areas is:

[0075] For any suspected pollution area in the color image of the bottom surface of the ship, the similarity factor of the suspected pollution area is obtained by comparing the similarity of each main color in the suspected pollution area;

[0076] Any other suspected contaminated area in the color image of the bottom surface of the ship is recorded as a reference suspected contaminated area; the Euclidean distance between the cluster center of the reference suspected contaminated area and the cluster center of the suspected contaminated area is recorded as the target distance of the reference suspected contaminated area; all the reference suspected contaminated areas in the color image of the bottom surface of the ship are sorted from large to small according to the target distance to obtain a reference area sequence of the suspected contaminated area; the first 8 reference suspected contaminated areas in the reference area sequence of the suspected contaminated area are all used as target suspected contaminated areas of the suspected contaminated area; the mean of the similarity factors of all target suspected contaminated areas of the suspected contaminated area is used as the texture complexity of the suspected contaminated area.

[0077] The specific formula is:

[0078]

[0079] Where SA represents the texture complexity of the suspected contaminated area; N represents the number of all target suspected contaminated areas in the suspected contaminated area; AS i Represents the similarity factor of the i-th target suspected contaminated area of ​​the suspected contaminated area.

[0080] Preferably, in some implementations of the embodiments of the present invention, the specific method for obtaining the similarity factor of the suspected contaminated area is:

[0081] Combining any two main colors in the suspected contaminated area as a main color combination of the suspected contaminated area;

[0082] For any main color combination in the suspected contaminated area, the degree of overlap of the main color combination is obtained according to the spatial average distance of the main colors in the main color combination;

[0083] The ratio of the overlap of the main color combination to the color distance between the two main colors in the main color combination is taken as the similarity of the main color combination; the average of the similarities of all the main color combinations in the suspected contaminated area is taken as the similarity factor of the suspected contaminated area;

[0084] The specific formula is:

[0085]

[0086] Where AS represents the similarity factor of the suspected contaminated area; n represents the number of all main color combinations of the suspected contaminated area; j Indicates the overlap of the jth main color combination of the suspected contaminated area; Dif j Represents the color distance between the two main colors in the jth main color combination of the suspected contaminated area.

[0087] Preferably, in some implementations of the embodiments of the present invention, the method for obtaining the degree of overlap of the main color combination includes:

[0088] The Euclidean distance between the spatial centers of the two main colors in the main color combination on the suspected pollution area is recorded as the spatial distance of the main color combination; the sum of the spatial average distances of the two main colors in the main color combination is recorded as the distance sum value; the ratio of the spatial distance of the main color combination to the distance sum value is recorded as the distance ratio; the difference between 1 and the distance ratio is taken as the overlap degree of the main color combination.

[0089] The specific formula is:

[0090]

[0091] Where O represents the overlap of the main color combination; D represents the Euclidean distance between the spatial centers of the two main colors in the main color combination on the suspected contaminated area; R1 represents the spatial average distance of the first main color in the main color combination; R2 represents the spatial average distance of the second main color in the main color combination.

[0092] At this point, the texture complexity of each suspected pollution area is obtained.

[0093] 3. Obtain all polluted areas in the color image of the bottom surface of the ship.

[0094] It should be noted that the greater the color complexity, the greater the possibility that the suspected polluted area is a polluted area; the smaller the texture complexity, the greater the possibility that the suspected polluted area is a polluted area.

[0095] Preferably, in some implementations of the embodiments of the present invention, all suspected polluted areas in the color image of the bottom surface of the ship are screened according to the color complexity and texture complexity, and the specific method for obtaining all polluted areas is:

[0096] A threshold parameter T is preset, wherein this embodiment is described by taking T=0.7 as an example, and this embodiment is not specifically limited, wherein T is determined according to specific implementation conditions;

[0097] For any suspected pollution area in the color image of the bottom surface of the ship, the normalized value of the ratio of the color complexity of the suspected pollution area to the texture complexity of the suspected pollution area is used as the possible factor of the suspected pollution area;

[0098] If the possible factor of the suspected polluted area is greater than the threshold parameter T, the suspected polluted area is recorded as a polluted area.

[0099] The specific formula is:

[0100]

[0101] Where PA represents the possible factors of the suspected polluted area; P represents the color complexity of the suspected polluted area; SA represents the texture complexity of the suspected polluted area; norm() represents the linear normalization function.

[0102] At this point, all polluted areas in the color image of the ship bottom surface are obtained.

[0103] 4. Obtain the degree of enhancement to be achieved in each polluted area.

[0104] It should be noted that different polluted areas require different degrees of enhancement. The growth of marine polluting organisms on the bottom surface of the ship has obvious ductility and strong adsorption, that is, its distribution is basically not changed by the flow of water. Taking barnacles as an example, their distribution is uniform, that is, usually other similar polluted areas are evenly distributed around a polluted area. The distribution of rust and other dirt on the bottom surface of the ship is more random, and it is not the main target of underwater robots for cleaning the bottom of the ship. Therefore, it is necessary to enhance the above-mentioned barnacles and other polluted areas in the image to a greater extent, and to enhance similar rust areas to a smaller extent.

[0105] Preferably, in some implementations of the embodiments of the present invention, the specific method for obtaining the degree of enhancement of the polluted area according to the color complexity and the distribution of the polluted area and the surrounding polluted areas is:

[0106] For any polluted area in the color image of the bottom surface of the ship, any other polluted area in the color image of the bottom surface of the ship is recorded as a reference polluted area; the Euclidean distance between the cluster center of the reference polluted area and the cluster center of the polluted area is recorded as the target reference distance of the reference polluted area; all reference polluted areas in the color image of the bottom surface of the ship are sorted from large to small according to the target reference distance to obtain a reference polluted area sequence of the polluted area; the first 8 reference polluted areas in the reference polluted area sequence of the polluted area are all used as the target polluted areas of the polluted area;

[0107] The variance of the Euclidean distance between the centroid of the polluted area and the centroid of all target polluted areas is recorded as the first variance; the cumulative sum of the Euclidean distances between the centroid of the polluted area and the centroid of all target polluted areas is recorded as the first cumulative sum; the inverse proportional value of the product of the first variance and the first cumulative sum is recorded as the first inverse proportional value; the product of the color complexity of the polluted area and the first inverse proportional value is taken as the degree to be enhanced of the polluted area.

[0108] The specific formula is:

[0109]

[0110] Where α represents the degree of enhancement to be performed in the polluted area; P represents the color complexity of the polluted area; σ represents the variance of the Euclidean distance between the centroid of the polluted area and the centroid of all target polluted areas; K represents the number of all target polluted areas in the polluted area; d k It represents the Euclidean distance between the centroid of the polluted area and the centroid of the kth target polluted area; exp() represents an exponential function with a natural constant as the base. The embodiment adopts the exp(-x) model to present the inverse proportional relationship. x is the input of the model. The implementer can select the inverse proportional function according to the actual situation.

[0111] It should be noted that It represents the external uniformity of the polluted area in the image. The larger the value, the more uneven the spatial distribution of the polluted area and its adjacent polluted areas, and the more likely it is that the polluted area is not a marine pollutant. The larger the value of color complexity, the more the polluted area needs to be strengthened.

[0112] So far, the degree to be enhanced of each polluted area is obtained through the above method.

[0113] Step S004: enhancing the color image of the bottom surface of the ship based on the degree to be enhanced, and obtaining an enhanced bottom surface image of the ship.

[0114] Preferably, in some implementations of the embodiments of the present invention, the color image of the bottom surface of the ship is enhanced based on the degree to be enhanced, and the specific method for obtaining the enhanced bottom surface image of the ship is:

[0115] The color image of the bottom surface of the ship is converted into the HSI color model, and the brightness and saturation of the polluted area in the image are enhanced to obtain an enhanced image of the bottom surface of the ship;

[0116] Taking any polluted area as an example, the brightness component and saturation component of each pixel therein are multiplied by the degree to be enhanced corresponding to the polluted area to obtain the enhanced brightness and saturation of each pixel.

[0117] See also Figure 2 , which shows a characteristic relationship flow chart of an underwater image enhancement method for underwater robot cleaning.

[0118] At this point, this embodiment is completed.

[0119] 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 underwater image enhancement method for underwater robot cleaning, characterized in that: The method comprises the following steps: Acquire multiple continuous color images of the bottom surface of the ship; By clustering multiple color images of the bottom surface of the ship, all suspected pollution areas in the color images of the bottom surface of the ship are obtained; Performing color quantization on the color image of the bottom surface of the ship to obtain a main color image of the color image of the bottom surface of the ship; the main color image includes several main colors; obtaining the color complexity of each suspected polluted area according to the number of main colors in each suspected polluted area and the spatial distribution of the main colors; obtaining the texture complexity of the suspected polluted area according to the similarity between the suspected polluted area and the main colors in the surrounding suspected polluted areas; screening all suspected polluted areas in the color image of the bottom surface of the ship according to the color complexity and the texture complexity to obtain all polluted areas; obtaining the degree of enhancement to be obtained in the polluted area according to the color complexity and the distribution of the polluted area and the surrounding polluted areas; The color image of the bottom surface of the ship is enhanced based on the degree to be enhanced to obtain an enhanced image of the bottom surface of the ship; The method of obtaining the degree of enhancement of the polluted area by the color complexity and the distribution of the polluted area and the surrounding polluted areas includes: For any polluted area in the color image of the bottom surface of the ship, any other polluted area in the color image of the bottom surface of the ship is recorded as a reference polluted area; the Euclidean distance between the cluster center of the reference polluted area and the cluster center of the polluted area is recorded as the target reference distance of the reference polluted area; all reference polluted areas in the color image of the bottom surface of the ship are sorted from large to small according to the target reference distance to obtain a reference polluted area sequence of the polluted area; the first 8 reference polluted areas in the reference polluted area sequence of the polluted area are all used as the target polluted areas of the polluted area; The variance of the Euclidean distance between the centroid of the polluted area and the centroid of all target polluted areas is recorded as the first variance; the cumulative sum of the Euclidean distances between the centroid of the polluted area and the centroid of all target polluted areas is recorded as the first cumulative sum; the inverse proportional value of the product of the first variance and the first cumulative sum is recorded as the first inverse proportional value; the product of the color complexity of the polluted area and the first inverse proportional value is taken as the degree to be enhanced of the polluted area.

2. The underwater image enhancement method for underwater robot cleaning according to claim 1, characterized in that: The method of clustering multiple color images of the bottom surface of the ship to obtain all suspected pollution areas in the color images of the bottom surface of the ship includes: For any color image of the bottom surface of a ship, the number of cluster centers and the initial cluster center positions of the color image of the bottom surface of the ship are obtained by using the elbow method; the number of cluster centers and the initial cluster center positions of the color image of the bottom surface of the ship are input into the K-means clustering algorithm to cluster the color image of the bottom surface of the ship to obtain a number of cluster clusters; For any cluster, in any ship bottom surface color image, the Euclidean distance between the cluster center of the cluster and the center position of the ship bottom surface color image is recorded as the first distance of the cluster in the ship bottom surface color image; if the first distances of the cluster in all ship bottom surface color images are the same, the area to which the cluster belongs is regarded as a suspected pollution area.

3. The underwater image enhancement method for underwater robot cleaning according to claim 1, characterized in that: The method of obtaining the color complexity of each suspected polluted area according to the number of main colors in each suspected polluted area and the spatial distribution of the main colors includes: For any suspected pollution area in the color image of the bottom surface of the ship, the spatial average distance and spatial distribution value of each main color are obtained according to the spatial distribution of each main color in the suspected pollution area in the main color image; The sum of the normalized value of the mean of the spatial distribution values ​​of all main colors in the suspected pollution area and 1 is recorded as the first sum value; The product of the number of all main colors in the suspected pollution area, the cumulative sum of the spatial average distances of all main colors, and the first sum is taken as the color complexity of the suspected pollution area.

4. The underwater image enhancement method for underwater robot cleaning according to claim 3, characterized in that: The specific method of obtaining the spatial average distance and spatial distribution value of each primary color includes: For any main color in the suspected pollution area, the average of the Euclidean distances between all the pixels of the main color and the centroid of the suspected pollution area is taken as the spatial average distance of the main color; The product of the square value of the spatial average distance of the main color and π is recorded as the first product; the ratio of the first product to the area of ​​the suspected pollution area is taken as the spatial distribution value of the main color.

5. The underwater image enhancement method for underwater robot cleaning according to claim 1, characterized in that: The specific method of obtaining the texture complexity of the suspected polluted area according to the similarity between the main colors of the suspected polluted area and the surrounding suspected polluted areas is as follows: For any suspected pollution area in the color image of the bottom surface of the ship, the similarity factor of the suspected pollution area is obtained by comparing the similarity of each main color in the suspected pollution area; Any other suspected pollution area in the color image of the bottom surface of the ship is recorded as a reference suspected pollution area; The Euclidean distance between the cluster center of the reference suspected pollution area and the cluster center of the suspected pollution area is recorded as the target distance of the reference suspected pollution area; all the reference suspected pollution areas in the color image of the bottom surface are sorted from large to small according to the target distance to obtain the reference area sequence of the suspected pollution area; the first 8 reference suspected pollution areas in the reference area sequence of the suspected pollution area are taken as the target suspected pollution areas of the suspected pollution area; the mean of the similarity factors of all the target suspected pollution areas of the suspected pollution area is taken as the texture complexity of the suspected pollution area.

6. The underwater image enhancement method for underwater robot cleaning according to claim 5, characterized in that: The specific method of obtaining the similarity factor of the suspected contaminated area includes: Combining any two main colors in the suspected contaminated area as a main color combination of the suspected contaminated area; For any main color combination in the suspected contaminated area, the degree of overlap of the main color combination is obtained according to the spatial average distance of the main colors in the main color combination; The ratio of the overlap of the main color combination to the color distance between the two main colors in the main color combination is taken as the similarity of the main color combination; the average of the similarities of all main color combinations in the suspected contaminated area is taken as the similarity factor of the suspected contaminated area.

7. The underwater image enhancement method for underwater robot cleaning according to claim 6, characterized in that: The specific method of obtaining the overlap degree of the main color combination includes: The Euclidean distance between the spatial centers of the two main colors in the main color combination on the suspected pollution area is recorded as the spatial distance of the main color combination; the sum of the spatial average distances of the two main colors in the main color combination is recorded as the distance sum value; the ratio of the spatial distance of the main color combination to the distance sum value is recorded as the distance ratio; the difference between 1 and the distance ratio is taken as the overlap degree of the main color combination.

8. The underwater image enhancement method for underwater robot cleaning according to claim 1, characterized in that: The method of screening all suspected polluted areas in the color image of the bottom surface of the ship according to the color complexity and texture complexity to obtain all polluted areas includes: A threshold parameter T is preset, and for any suspected pollution area in the color image of the bottom surface of the ship, possible factors of the suspected pollution area are obtained according to the color complexity and texture complexity; If the possible factor of the suspected polluted area is greater than the threshold parameter T, the suspected polluted area is recorded as a polluted area.

9. The underwater image enhancement method for underwater robot cleaning according to claim 8, characterized in that: The specific method of obtaining possible factors of the suspected contaminated area includes: The normalized value of the ratio of the color complexity of the suspected polluted area to the texture complexity of the suspected polluted area is used as the possible factor of the suspected polluted area.

Images