A method for monitoring pollutants in aquaculture areas

Abstract

The present invention relates to the technical field of water surface pollutant monitoring, and specifically to a pollutant monitoring method for an aquaculture area, comprising: collecting a mixed image of water blooms and fallen leaves, graying the mixed image of water blooms and fallen leaves to obtain a mixed grayscale image of water blooms and fallen leaves; performing threshold segmentation on the mixed grayscale image of water blooms and fallen leaves according to a preset fallen leaf threshold to distinguish fallen leaf areas and suspected water bloom areas; performing threshold segmentation on the suspected water bloom areas according to a preset water bloom threshold to obtain water bloom areas; obtaining the ratio of the area of ​​the water bloom area to the area of ​​the mixed image of water blooms and fallen leaves, and judging the degree of water bloom pollution and countermeasures according to the ratio. By adopting the present invention, the degree of water bloom pollution can be obtained according to the mixed image of water blooms and fallen leaves, and corresponding measures can be taken in time to deal with these water blooms according to the degree of water bloom pollution.

Description

Technical Field

[0001] The present invention relates to the technical field of water surface pollutant monitoring, and in particular to a method for monitoring pollutants in an aquaculture area. Background Art

[0002] Intelligent monitoring systems have shown significant advantages and broad application prospects in monitoring water surface pollutants, especially the algal bloom problem faced in aquaculture. As a common pollutant in aquaculture, the large-scale outbreak of algal bloom will seriously deteriorate water quality, reduce the amount of dissolved oxygen in the water, and thus break the ecological balance of aquaculture, and such problems are often difficult to effectively control. Therefore, it is particularly important to conduct timely and accurate detection of algal blooms in aquaculture areas. The intelligent monitoring system mainly uses camera detection technology to monitor the algal bloom situation on the local water surface. This technology can rely on ground-mounted cameras or cameras mounted on drones to operate, combined with image recognition and analysis technology to comprehensively monitor the algal bloom situation in the water body. The advantage of camera detection technology is that it can capture high-resolution images, which is very suitable for detailed monitoring of local areas.

[0003] However, there are also certain challenges in identifying algal blooms in local aquaculture areas using existing threshold segmentation and edge detection technologies. Specifically, since the algal bloom area and the fallen leaf area are similar in features and have similar grayscale values, the two are easily mixed together in the image, making it difficult for the system to accurately and completely extract the algal bloom area, and thus unable to obtain the degree of algal bloom pollution, making it impossible for relevant staff to take corresponding measures based on the degree of algal bloom pollution, which may cause serious losses to aquaculture in many aspects. Summary of the invention

[0004] In order to solve the technical problem that the prior art cannot accurately and completely extract the water bloom area from the mixed image of water bloom and fallen leaves, and thus cannot obtain the degree of water bloom pollution and take corresponding measures, the purpose of the present invention is to provide a method for monitoring pollutants in aquaculture areas. The technical scheme adopted is as follows:

[0005] Collecting a mixed image of water blooms and fallen leaves, and graying the mixed image of water blooms and fallen leaves to obtain a mixed grayscale image of water blooms and fallen leaves;

[0006] Performing threshold segmentation on the mixed grayscale image of water bloom and fallen leaves according to a preset fallen leaf threshold to distinguish fallen leaf areas and suspected water bloom areas;

[0007] Performing threshold segmentation on the suspected algal bloom area according to a preset algal bloom threshold to obtain an algal bloom area;

[0008] The ratio of the area of ​​the water bloom region to the area of ​​the mixed image of water bloom and fallen leaves is obtained, and the degree of water bloom pollution and the corresponding countermeasures are determined according to the ratio.

[0009] Furthermore, the method further comprises:

[0010] Segmenting the mixed grayscale image of water bloom and fallen leaves from small to large according to a segmentation threshold, to obtain a preset number of segmented images of water bloom and fallen leaves, wherein a first value range of the segmentation threshold is 0 to 225;

[0011] Denoising the water bloom and fallen leaves segmented image, so that a value range of a segmentation threshold corresponding to the denoised water bloom and fallen leaves segmented image is narrowed from the first value range to a second value range;

[0012] Performing Otsu threshold segmentation on the mixed grayscale image of water bloom and fallen leaves within the second value range to obtain an optimal segmentation threshold;

[0013] A first threshold range and a second threshold range are divided according to the optimal segmentation threshold. In the second value range, the segmentation threshold before the optimal segmentation threshold is the first threshold range, and the segmentation threshold after the optimal segmentation threshold is the second threshold range.

[0014] Furthermore, the process of obtaining the leaf drop threshold includes:

[0015] Performing threshold segmentation on the mixed grayscale image of water bloom and fallen leaves within the first threshold range and extracting a first initial threshold segmentation image corresponding to the pixel point where fallen leaves first appear, wherein the segmentation threshold corresponding to the first initial threshold segmentation image is the first initial segmentation threshold;

[0016] Locating the fallen leaf pixels in the first initial threshold segmentation image, clustering the fallen leaf pixels, determining the central pixel of each class and establishing a window so that each window contains all the fallen leaf pixels in the corresponding class;

[0017] The segmentation threshold of the mixed grayscale image of water blooms and fallen leaves is increased from the first initial segmentation threshold by a preset threshold value, and the number of newly added pixels in any of the windows and the average distance between the newly added pixels and the central pixel are obtained;

[0018] Establishing a first curve graph with the number of newly added pixels as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis;

[0019] Establishing a second curve graph with the average distance as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis;

[0020] Obtaining points reaching a first overall significance level on the first curve graph and the second curve graph, and obtaining segmentation thresholds corresponding to the points;

[0021] Arrange the segmentation thresholds from large to small, and calculate the difference between adjacent segmentation thresholds;

[0022] It is determined whether the value obtained by adding the preset first value to the average value of the difference and then taking the inverse value is greater than the preset first threshold. If so, the maximum segmentation threshold is the fallen leaf threshold.

[0023] Further, obtaining points reaching a first overall significance level on the first curve graph and the second curve graph includes:

[0024] Obtaining a first significance level of each point on the first curve graph with respect to a mutation feature and a second significance level of each point on the second curve graph with respect to an aggregation feature, wherein the first significance level corresponds to the second significance level in one-to-one correspondence;

[0025] It is determined whether the product of the first significance level and the corresponding second significance level is greater than a preset first overall significance threshold; if so, the point reaches the first overall significance level.

[0026] Furthermore, the process of obtaining the first significance level includes:

[0027] Obtaining a first slope corresponding to the cth point on the first curve graph, a first maximum slope and a first minimum slope corresponding to the points on the first curve graph, wherein c is a positive integer in the range of 1-21;

[0028] A value obtained by subtracting the first minimum slope from the first slope and dividing the result by the first maximum slope is the first significance level of the c-th point with respect to the mutation feature.

[0029] Furthermore, the process of obtaining the second significance level includes:

[0030] Obtaining a second slope corresponding to the c-th point on the second curve graph, where c is a positive integer in the range of 1-21;

[0031] The reciprocal value of the second slope plus the preset second value is calculated as the second significance level of the c-th point with respect to the clustering feature.

[0032] Further, performing threshold segmentation on the suspected algal bloom area according to a preset algal bloom threshold to obtain the algal bloom area includes:

[0033] Performing threshold segmentation on the suspected algal bloom area within the second threshold range and extracting a second initial threshold segmentation image corresponding to the initially appearing pixel point, wherein the segmentation threshold corresponding to the second initial threshold segmentation image is the second initial segmentation threshold;

[0034] Locating the pixel points in the second initial threshold segmented image, clustering the pixel points, determining the central pixel point of each class and establishing a window so that each window contains all the pixel points in the corresponding class;

[0035] The segmentation threshold of the threshold segmented image is increased sequentially by a preset threshold value, and the number of newly added pixel points in any of the windows and the average distance between the newly added pixel points and the central pixel point are obtained;

[0036] Establishing a third curve graph with the number of newly added pixels as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis;

[0037] Establishing a fourth curve graph with the average distance as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis;

[0038] Obtaining points reaching a second overall significance level on the third curve graph and the fourth curve graph, and obtaining segmentation thresholds corresponding to the points;

[0039] The segmentation threshold is subtracted from a preset third value and the absolute value is obtained, and the number of channels after adding a preset fourth value to the absolute value is obtained to obtain the possibility;

[0040] The segmentation threshold corresponding to the one with the greatest possibility is the water bloom threshold.

[0041] Further, obtaining the points reaching the second overall significance level on the third curve graph and the fourth curve graph includes:

[0042] Obtaining a third significance level of each point on the third curve graph with respect to the mutation feature and a fourth significance level of each point on the fourth curve graph with respect to the aggregation feature, wherein the third significance level corresponds to the fourth significance level in a one-to-one manner;

[0043] It is determined whether the product of the third significance level and the corresponding fourth significance level is greater than a preset second overall significance threshold; if so, the point reaches the second overall significance level.

[0044] Furthermore, the process of obtaining the third significance level includes:

[0045] Obtaining a first slope corresponding to the cth point on the third curve graph, a second maximum slope and a second minimum slope corresponding to the point on the third curve graph, wherein c is a positive integer in the range of 1-15;

[0046] The value obtained by subtracting the second minimum slope from the third slope and dividing the result by the second maximum slope is the third significance level of the c-th point with respect to the mutation feature.

[0047] Furthermore, judging the degree of algal bloom pollution and the corresponding measures according to the ratio include:

[0048] When the ratio is less than the first pollution threshold, the water bloom pollution level is light, and a first response measure is taken, wherein the first response measure is to add a microbial agent to the water bloom pollution area to prevent the growth of water blooms;

[0049] When the ratio is greater than or equal to the first pollution threshold and less than or equal to the second pollution threshold, the degree of water bloom pollution is moderate, and a second response measure is taken, wherein the second response measure is to set up a fence and add an aerator to the water bloom pollution area;

[0050] When the ratio is greater than the second pollution threshold, the water bloom pollution level is severe, and a third response measure is taken, which is to spray flocculants on the water bloom pollution area and add a preset amount of activated carbon adsorbent.

[0051] The present invention has the following beneficial effects:

[0052] In the present invention, first, a mixed image of water bloom and fallen leaves is collected, and the mixed image of water bloom and fallen leaves is grayed to obtain a mixed gray image of water bloom and fallen leaves. The mixed gray image of water bloom and fallen leaves is the image basis for subsequent analysis. Secondly, the mixed gray image of water bloom and fallen leaves is threshold segmented according to a preset fallen leaf threshold to distinguish the fallen leaf area and the suspected water bloom area. Here, the fallen leaf area and the suspected water bloom area are first divided, and the suspected water bloom area needs to be further determined whether it is a water bloom area. Furthermore, the suspected water bloom area is threshold segmented according to a preset water bloom threshold to obtain the water bloom area. Here, the water bloom area is determined by the suspected water bloom area. Finally, the ratio of the area of ​​the water bloom area to the area of ​​the mixed image of water bloom and fallen leaves is obtained, and the degree of water bloom pollution and the countermeasures are judged according to the ratio. Here, the degree of water bloom pollution is judged according to the water bloom pollution degree, and corresponding countermeasures are taken according to the water bloom pollution degree, so that the water bloom pollution situation is alleviated or solved. By adopting the present invention, the degree of water bloom pollution can be obtained according to the mixed image of water bloom and fallen leaves, and corresponding measures can be taken in time to deal with these water blooms according to the degree of water bloom pollution, so as to avoid excessive water blooms causing water quality deterioration, aquatic biological diseases and deaths, ecosystem degradation and biodiversity reduction, economic losses and other losses in aquaculture. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are 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.

[0054] Figure 1 A flow chart of a method for monitoring pollutants in an aquaculture area provided by the first embodiment of the present invention;

[0055] Figure 2 Another flow chart of a method for monitoring pollutants in an aquaculture area provided by the second embodiment of the present invention.

[0056] Figure 3 A flowchart of a process for obtaining a leaf fall threshold value provided in the third embodiment of the present invention;

[0057] Figure 4 A flowchart for obtaining points reaching a first overall significance level on the first curve graph and the second curve graph provided in the fourth embodiment of the present invention;

[0058] Figure 5 A flowchart of a process for obtaining a first saliency level provided by a fifth embodiment of the present invention;

[0059] Figure 6 A flowchart of a process for obtaining a second significance level provided by a sixth embodiment of the present invention;

[0060] Figure 7 A flowchart of a process for obtaining an algae bloom threshold value provided by the seventh embodiment of the present invention;

[0061] Figure 8 A flowchart for obtaining points reaching a second overall significance level on the third curve graph and the fourth curve graph provided in the eighth embodiment of the present invention;

[0062] Fig. 9 A flowchart of a process for obtaining a third saliency level provided by a ninth embodiment of the present invention;

[0063] Fig.10 This is a flow chart of a process for obtaining a fourth saliency level provided in the tenth embodiment of the present invention. DETAILED DESCRIPTION

[0064] 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 pollutant monitoring method for aquaculture areas proposed by the present invention, its specific implementation method, structure, characteristics 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.

[0065] 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.

[0066] The specific scheme of the pollutant monitoring method for aquaculture areas provided by the present invention is described in detail below with reference to the accompanying drawings.

[0067] See also Figure 1 , which shows a flow chart of a pollutant monitoring method for an aquaculture area provided by a first embodiment of the present invention, the method comprising:

[0068] S101. Collecting a mixed image of water blooms and fallen leaves, and gray-scaling the mixed image of water blooms and fallen leaves to obtain a mixed grayscale image of water blooms and fallen leaves.

[0069] The camera captures the light reflected from the water surface in the local area of ​​the small-scale aquaculture area through the lens, and the light is focused by the lens and projected onto the image sensor (such as CCD or CMOS). The image sensor converts the optical image into an electrical signal, which is then converted into a digital signal through an analog-to-digital converter (ADC), and finally processed by a digital signal processing chip (DSP) to generate a mixed image of water blooms and fallen leaves.

[0070] In the mixed image of water bloom and fallen leaves, the water bloom and fallen leaves are closely connected and have similar colors. In the mixed image of water bloom and fallen leaves, fallen leaves are piled up, the water bloom area is formed by a large number of algae, the leaf edge features in the fallen leaf area are not significant enough, and the water bloom area does not have fixed shape features. The water bloom area cannot be accurately extracted using existing edge detection technology. However, through further analysis of the mixed image of water bloom and fallen leaves, it is found that although the water bloom and fallen leaves are similar in color, the water bloom area is mostly yellow-green, and the fallen leaf area is brown or light yellow, that is, there is still a grayscale difference in the grayscale images of the two regions. Although single threshold segmentation cannot accurately extract the water bloom area, due to the grayscale difference between the two, there are different performances when different threshold segmentations are used, and this point can be used to distinguish them.

[0071] Gray-scaling the mixed image of water blooms and fallen leaves is a prior art and will not be described in detail here.

[0072] S102. Performing threshold segmentation on the mixed grayscale image of water bloom and fallen leaves according to a preset fallen leaf threshold to distinguish fallen leaf areas and suspected water bloom areas.

[0073] The process of obtaining the leaf-falling threshold will be described in detail in the third embodiment and will not be repeated here.

[0074] S103. Perform threshold segmentation on the suspected algal bloom area according to a preset algal bloom threshold to obtain an algal bloom area.

[0075] The process of obtaining the water bloom threshold will be described in detail in the seventh embodiment and will not be repeated here.

[0076] S104. Obtain a ratio of the area of ​​the water bloom region to the area of ​​the mixed image of water blooms and fallen leaves, and determine the degree of water bloom pollution and countermeasures based on the ratio.

[0077] Specifically, judging the degree of algal bloom pollution and the corresponding measures according to the ratio include:

[0078] When the ratio is less than the first pollution threshold, the water bloom pollution is mild, and the first response measure is taken. The first response measure is to add a microbial agent to the water bloom pollution area to prevent the growth of water blooms. The microbial agent may include: green ecological algae control agent or cyanobacteria ecological regulation bacteria. By adding microbial agents such as green ecological algae control agent or cyanobacteria ecological regulation bacteria, the water bloom growth chain is blocked and its photosynthesis is cut off. The first pollution threshold can be set independently, preferably 0.3.

[0079] When the ratio is greater than or equal to the first pollution threshold and less than or equal to the second pollution threshold, the algae pollution level is moderate, and the second response measure is taken, which is to set up fences and add aerators in the algae pollution area; fence facilities are set up in the water area where algae blooms are found, which can control the algae in a certain area to prevent it from spreading to the water intake, and the use of aerators and other methods can reduce and control the growth and spread of algae. The second pollution threshold can be set independently, preferably 0.8.

[0080] When the ratio is greater than the second pollution threshold, the water bloom pollution is severe, and a third response measure is taken, which is to spray a flocculant on the water bloom pollution area and add a preset amount of activated carbon adsorbent. The flocculant includes: polyaluminium chloride, polyaluminium chloride iron or chitosan modified kaolin, etc.

[0081] Figure 2 Another flow chart of a method for monitoring pollutants in an aquaculture area provided by the second embodiment of the present invention, the method further comprising:

[0082] S201. Segment the mixed grayscale image of water bloom and fallen leaves from small to large according to a segmentation threshold, to obtain a preset number of water bloom and fallen leaves segmentation images, wherein the first value range of the segmentation threshold is 0 to 225.

[0083] The mixed grayscale image of water bloom and fallen leaves is threshold segmented from small to large. The first value range is 0 to 225. Here, the corresponding segmented image can be extracted and analyzed every 5 thresholds on average, and 45 threshold segmented images are obtained. In each threshold segmented image, the area where the pixels greater than or equal to the segmentation threshold are located is the suspected water bloom area, and the area where the pixels less than the segmentation threshold are located is the fallen leaf area.

[0084] S202. De-noising the water bloom and fallen leaves segmented image, and reducing the value range of the segmentation threshold corresponding to the de-noised water bloom and fallen leaves segmented image from the first value range to the second value range.

[0085] The water bloom and fallen leaf segmentation images, i.e., 45 threshold segmentation images, are denoised, i.e., images in which no pixels exist due to a too small threshold and segmentation images before the threshold are removed, as well as images in which all areas are target pixels due to a too large threshold and segmentation images after the threshold are removed.

[0086] After denoising, 37 threshold segmentation images with segmentation thresholds ranging from 15 to 200 were obtained.

[0087] That is to say, the original first value range of 0 to 225 is reduced to the second value range of 15 to 200.

[0088] S203. Perform Otsu threshold segmentation on the mixed grayscale image of water blooms and fallen leaves within the second value range to obtain an optimal segmentation threshold.

[0089] By performing Otsu threshold segmentation on the mixed grayscale image of water bloom and fallen leaves, an Otsu threshold segmentation map with an optimal threshold of 123 is obtained.

[0090] S204. Divide the optimal segmentation threshold into a first threshold range and a second threshold range according to the optimal segmentation threshold. In the second value range, the segmentation threshold before the optimal segmentation threshold is the first threshold range, and the segmentation threshold after the optimal segmentation threshold is the second threshold range.

[0091] By observing these 37 threshold segmentation images with segmentation thresholds ranging from 15 to 200 and the segmentation thresholds of adjacent threshold segmentation images separated by 5, and combining the optimal segmentation threshold, we found that:

[0092] The threshold segmentation image with a segmentation threshold of 15 to 120 is segmented by increasing the segmentation threshold by 5. During the segmentation process, the pixel points of the corresponding fallen leaf area in the mixed grayscale image of water bloom fallen leaves show mutation characteristics and aggregation characteristics on multiple threshold segmentation images.

[0093] The segmentation threshold is 125 to 200, and the segmentation threshold is continuously increased in increments of 5 for segmentation. During the segmentation process, the pixels of the corresponding water bloom area in the mixed grayscale image of water bloom and fallen leaves show a slow growth trend at first, and after the segmentation threshold is 185, stronger mutation characteristics and aggregation characteristics begin to appear.

[0094] That is to say, according to the above characteristics, the optimal segmentation threshold can be divided into a first threshold range and a second threshold range. The segmentation threshold before the optimal segmentation threshold is the first threshold range, i.e., the low threshold range. Within the first threshold range, a certain fallen leaf threshold can be found to well distinguish the fallen leaf area. The segmentation threshold after the optimal segmentation threshold is the second threshold range, i.e., the high threshold range. Within the second threshold range, a certain algae bloom threshold can be found to well distinguish the algae bloom area.

[0095] In the mixed grayscale image of water bloom fallen leaves, the fallen leaves area is brown or light yellow, and the grayscale value is smaller than that of the water bloom area. When the mixed grayscale image of water bloom fallen leaves is segmented from small to large thresholds, the first pixels segmented out are some of the pixels in the fallen leaves area. In the mixed grayscale image of water bloom fallen leaves, the fallen leaves area is mostly brown or light yellow, so the grayscale value difference shown in the mixed grayscale image of water bloom fallen leaves, that is, the difference between the grayscale value of fallen leaves and the grayscale value of water bloom, will be stable in a small interval. When the mixed grayscale image of water bloom fallen leaves is segmented from small to large thresholds, when the segmentation threshold reaches this small interval, the pixels in the fallen leaves area in the mixed grayscale image of water bloom fallen leaves will show strong aggregation characteristics and mutation characteristics, that is, when the threshold is segmented from small to large, the pixels in the fallen leaves area have aggregation characteristics and mutation characteristics, and there is a special threshold interval that makes the aggregation characteristics and mutation characteristics more significant, then the possibility that the corresponding pixel point area segmented at the end of the special threshold interval is the fallen leaves area is greater. This feature will be used to preliminarily determine the fallen leaf area, and then determine the suspected algal bloom area outside the fallen leaf area.

[0096] Figure 3 A flowchart of a process for obtaining a leaf-falling threshold provided in the third embodiment of the present invention, wherein the process for obtaining the leaf-falling threshold includes:

[0097] S301. Perform threshold segmentation on the mixed grayscale image of water bloom and fallen leaves within the first threshold range and extract the first initial threshold segmentation image corresponding to the pixel point where the fallen leaves first appear. The segmentation threshold corresponding to the first initial threshold segmentation image is the first initial segmentation threshold.

[0098] Due to the different grayscale values ​​of the bloom area and the fallen leaf area, the threshold segmentation image presents different states when the segmentation threshold changes from small to large. In the early stage of threshold segmentation, the fallen leaf area does not appear at the beginning, but when the segmentation threshold reaches a certain threshold, that is, the first initial segmentation threshold, the pixels in the fallen leaf area will gradually appear. To study the changing characteristics of the pixels in the fallen leaf area, it is necessary to establish a window based on the area where the fallen leaf pixels begin to appear, and analyze based on the changing characteristics of the pixels in a single window.

[0099] S302. Locate the fallen leaf pixels in the first initial threshold segmentation image, cluster the fallen leaf pixels, determine the central pixel of each cluster and establish a window so that each window contains all the fallen leaf pixels in the corresponding cluster.

[0100] When the mixed grayscale image of water bloom and fallen leaves is segmented from small to large thresholds within the first threshold range, the threshold segmentation image where the fallen leaf pixels first appear, i.e., the first initial threshold segmentation image, is extracted, and all fallen leaf pixels segmented from the first initial threshold segmentation image are located. These fallen leaf pixels are clustered by maximum and minimum distance to obtain b regions, and b windows are established with the central pixels of these regions. The window size should contain all fallen leaf pixels in each class as much as possible, and the value of b is determined according to the actual situation.

[0101] It should be noted that, in this embodiment, the grayscale value of the fallen leaves region is lower than that of the water bloom region. Therefore, when the whole image is segmented by threshold, the first region segmented is the fallen leaves region.

[0102] The clustering and central pixel determination method is a prior art, and for example, a K-means algorithm may be used.

[0103] S303. The segmentation threshold of the mixed grayscale image of water blooms and fallen leaves is increased from the first initial segmentation threshold by a preset threshold number, and the number of newly added pixels in any of the windows and the average distance between the newly added pixels and the central pixel are obtained.

[0104] The preset threshold number can be set to 5.

[0105] Here, the change in the number of newly added pixels is used to reflect the mutation feature as the segmentation threshold increases with the preset threshold number in the same window. The distance relationship between the newly added pixels and the central pixel is used to reflect the aggregation feature as the segmentation threshold increases with the preset threshold number in the same window.

[0106] S304. Create a first curve graph with the number of newly added pixels as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis.

[0107] In the first curve graph, the number of newly added pixels can reflect the degree of increase. The steeper the curve in the first curve graph, the greater the degree of increase.

[0108] S305. Create a second curve graph with the average distance as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis.

[0109] In the second curve graph, the smaller the average distance between the newly added pixels and the central pixel is, the more significant the clustering feature of the newly added pixels in the window is.

[0110] S306. Obtain points reaching a first overall significance level on the first curve graph and the second curve graph, and obtain segmentation thresholds corresponding to the points.

[0111] The process of obtaining the points reaching the first overall significance level on the first curve graph and the second curve graph will be described in detail in the fourth embodiment and will not be repeated here.

[0112] S307. Arrange the segmentation thresholds from small to large, and calculate the difference between adjacent segmentation thresholds.

[0113] The difference can be expressed as: , the value range of j is from 2 to the number of segmentation thresholds, represents the jth segmentation threshold, Indicates The segmentation threshold.

[0114] S308. Determine whether the value obtained by adding the preset first value to the average value of the difference and then taking the inverse value is greater than the preset first threshold. If so, the maximum segmentation threshold is the fallen leaf threshold.

[0115] The concentration and continuity of these segmentation thresholds are used to determine the likelihood that these segmentation thresholds are a certain special threshold interval mentioned in this article, thereby indicating the likelihood that the pixel point area segmented by the end threshold of the certain special threshold interval is a fallen leaf area.

[0116] The average value of the difference plus the preset first value and then the reciprocal value can be expressed as:

[0117] ;

[0118] Among them, the is the average value of the difference, H is the number of the segmentation thresholds, is the first value, preferably 1, in order to prevent the denominator from being 0, The value is equal to the average value of the difference plus the preset first value and then the inverse value, which indicates the possibility that the entire threshold interval composed of these segmentation thresholds is a certain special threshold interval. Here, the first threshold is set to 0.85 (empirical value). If the value is greater than 0.85, the entire threshold interval composed of these segmentation thresholds is considered to be a special threshold interval. Then the pixel area segmented by the end threshold of the special threshold interval, that is, the largest segmentation threshold, is the fallen leaf area. That is, the fallen leaf threshold is used to perform threshold segmentation on the image. The part less than the fallen leaf threshold belongs to the fallen leaf area, and the part greater than or equal to the fallen leaf threshold belongs to the non-fallen leaf area, that is, the suspected algal bloom area.

[0119] Then, the area segmented by the terminal threshold in the entire special threshold interval composed of these segmentation thresholds in the window area is extracted, and the area is the fallen leaf area.

[0120] Finally, based on the fallen leaf area extracted in the window area, all non-fallen leaf areas in the window are regarded as suspected water bloom areas.

[0121] Each window can extract the fallen leaf area according to the above process, so as to obtain the fallen leaf area in the water bloom and fallen leaf mixed image.

[0122] Figure 4 A flowchart of obtaining points reaching a first overall significance level on the first curve graph and the second curve graph provided in the fourth embodiment of the present invention, wherein obtaining points reaching a first overall significance level on the first curve graph and the second curve graph includes:

[0123] S401. Obtain a first significance level of each point on the first curve graph with respect to a mutation feature and a second significance level of each point on the second curve graph with respect to an aggregation feature, wherein the first significance level corresponds to the second significance level in a one-to-one manner.

[0124] The process of obtaining the first saliency level will be described in detail in the fifth embodiment and will not be repeated here.

[0125] The process of obtaining the second saliency level will be described in detail in the sixth embodiment and will not be repeated here.

[0126] The first significance level is the first significance level of a point on the first curve graph with respect to the mutation feature, and the second significance level is the second significance level of a point on the second curve graph with respect to the aggregation feature. When the horizontal coordinates of a point on the first curve graph and a point on the second curve graph are the same, the first significance level corresponds to the second significance level one by one.

[0127] S402. Determine whether the product of the first significance level and the corresponding second significance level is greater than a preset first overall significance threshold; if so, the point reaches the first overall significance level.

[0128] The product of the first significance level and the corresponding second significance level can be expressed as:

[0129] ;

[0130] Among them, the represents the first significance level, the represents the second level of significance, represents the product of the first significance level and the corresponding second significance level. The first overall significance threshold can be set to 0.8 (empirical value). If it is greater than 0.8, the point reaches the first overall significance level.

[0131] Figure 5 This is a flow chart of a process for obtaining a first saliency level provided in a fifth embodiment of the present invention. The process for obtaining the first saliency level includes:

[0132] S501. Obtain a first slope corresponding to the cth point on the first curve graph, a first maximum slope and a first minimum slope corresponding to the points on the first curve graph, wherein c is a positive integer in the range of 1-21.

[0133] The first slope can be expressed as:

[0134] ;

[0135] Among them, the is the number of newly added pixels corresponding to the c+1th point. is the number of newly added pixel points corresponding to the c-1th point, d represents the preset threshold number, and k represents the first slope.

[0136] According to the above process of obtaining the first slope, the slopes corresponding to the various points on the first curve graph can be obtained by analogy, wherein the one with the largest slope is the first maximum slope, and the one with the smallest slope is the first minimum slope.

[0137] S502. Subtract the first minimum slope from the first slope and divide the result by the first maximum slope to obtain the first significance level of the c-th point with respect to the mutation feature.

[0138] The first significance of the c-th point with respect to the mutation feature can be expressed as:

[0139] ;

[0140] Among them, the represents the first slope, the represents the first minimum slope, the represents the first maximum slope, the is the first significance degree of the c-th point with respect to the mutation feature.

[0141] Figure 6 This is a flow chart of a process for obtaining a second saliency level provided in a fifth embodiment of the present invention. The process for obtaining the second saliency level includes:

[0142] S601. Obtain a second slope corresponding to the cth point on the second curve graph, where c is a positive integer ranging from 1 to 21.

[0143] The second slope can be expressed as:

[0144] ;

[0145] Among them, the is the average distance corresponding to the c+1th point, is the average distance corresponding to the c-1th point, d represents the preset threshold value, is the second slope.

[0146] S602. Add a preset second value to the second slope and take the reciprocal value as the second significance level of the c-th point with respect to the clustering feature.

[0147] The smaller the average distance between the newly added pixels and the central pixel, the more significant the clustering feature of the newly added pixels in the window. The smaller it is, the smaller the change in the average distance is, which means that the clustering characteristics of the newly added pixels are more significant, so the second significance level is inversely proportional to the second slope.

[0148] The second significance level can be expressed as:

[0149] ;

[0150] Among them, the For the second level of significance, the is the second slope, the For the second value, it is preferably 1 to prevent the denominator from being 0.

[0151] Figure 7 The flowchart of the process of obtaining the water bloom threshold provided in the seventh embodiment of the present invention, wherein the suspected water bloom area is segmented according to the preset water bloom threshold to obtain the water bloom area includes:

[0152] S701. Perform threshold segmentation on the suspected water bloom area within the second threshold range and extract a second initial threshold segmentation image corresponding to the first appearing pixel point, wherein the segmentation threshold corresponding to the second initial threshold segmentation image is the second initial segmentation threshold.

[0153] To study the changing characteristics of pixel points in the algal bloom area, it is necessary to establish a window based on the algal bloom area where the algal bloom has already begun to appear, and perform analysis based on the changing characteristics of pixel points in a single window.

[0154] S702. Locate the pixel points in the second initial threshold segmentation image, cluster the pixel points, determine the central pixel point of each cluster and establish a window so that each window contains all the pixel points in the corresponding cluster.

[0155] When the mixed grayscale image of water bloom and fallen leaves is segmented from small to large thresholds within the second threshold range, the threshold segmentation image of the first pixel point is extracted, that is, the second initial threshold segmentation image, and all the pixel points segmented from the second initial threshold segmentation image are located. The maximum and minimum distance clustering is performed on these pixel points to obtain regions and establish a The window size should contain all pixels in each class as much as possible. The value is determined according to the actual situation.

[0156] The clustering and central pixel determination method is a prior art, and for example, a K-means algorithm may be used.

[0157] S703. The segmentation threshold of the threshold segmented image is increased by a preset threshold value in sequence, and the number of newly added pixels in any of the windows and the average distance between the newly added pixels and the central pixel are obtained.

[0158] The preset threshold number can be set to 5.

[0159] Here, the change in the number of newly added pixels is used to reflect the mutation feature as the segmentation threshold increases with the preset threshold number in the same window. The distance relationship between the newly added pixels and the central pixel is used to reflect the aggregation feature as the segmentation threshold increases with the preset threshold number in the same window.

[0160] S704. Create a third curve graph with the number of newly added pixels as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis.

[0161] In the third curve graph, the number of newly added pixels can reflect the degree of increase. The steeper the curve in the third curve graph, the greater the degree of increase.

[0162] S705. Create a fourth curve graph with the average distance as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis.

[0163] In the fourth curve graph, the smaller the average distance between the newly added pixel points and the central pixel point is, the more significant the clustering feature of the newly added pixel points in the window is.

[0164] S706. Obtain points reaching a second overall significance level on the third curve graph and the fourth curve graph, and obtain segmentation thresholds corresponding to the points.

[0165] The process of obtaining the points reaching the first overall significance level on the third curve graph and the fourth curve graph will be described in detail in the eighth embodiment and will not be repeated here.

[0166] S707. Subtract the preset third value from the segmentation threshold and take the absolute value, and add the preset fourth value to the absolute value to obtain the number of channels to obtain the possibility.

[0167] It can be seen that the mutation and aggregation characteristics of the algal bloom area are most significant when the segmentation threshold is the third value. The possibility of the area being an algal bloom area is judged by the proximity between the calculated interval of significant mutation and aggregation of the algal bloom area and the third value.

[0168] The possibility can be expressed as:

[0169] ;

[0170] Among them, the is the segmentation threshold, is the third value, preferably 185, is the fourth value, preferably 1, to prevent the denominator from being 0, and L is the possibility.

[0171] S708. The segmentation threshold corresponding to the one with the greatest possibility is the water bloom threshold. The suspected water bloom area is segmented using the water bloom threshold, and the pixel area greater than or equal to the water bloom threshold is the water bloom area.

[0172] Figure 8 The flowchart of obtaining the points reaching the second overall significance level on the third curve graph and the fourth curve graph provided in the eighth embodiment of the present invention, wherein obtaining the points reaching the second overall significance level on the third curve graph and the fourth curve graph comprises:

[0173] S801. Obtain a third significance level of each point on the third curve graph with respect to a mutation feature and a fourth significance level of each point on the fourth curve graph with respect to an aggregation feature, wherein the third significance level corresponds to the fourth significance level one by one.

[0174] The process of obtaining the third saliency level will be described in detail in the ninth embodiment and will not be repeated here.

[0175] The process of obtaining the fourth saliency level will be described in detail in the tenth embodiment and will not be repeated here.

[0176] The third significance level is the third significance level of a point on the third curve graph with respect to the mutation feature, and the fourth significance level is the fourth significance level of a point on the fourth curve graph with respect to the aggregation feature. When the horizontal coordinates of a point on the third curve and a point on the fourth curve graph are the same, the third significance level corresponds one-to-one to the fourth significance level.

[0177] The third significance level is the third significance level of a point on the third curve graph with respect to the mutation feature, and the fourth significance level is the fourth significance level of a point on the fourth curve graph with respect to the aggregation feature. When the horizontal coordinates of a point on the third curve graph and a point on the fourth curve graph are the same, the third significance level corresponds one-to-one to the fourth significance level.

[0178] S802. Determine whether the product of the third significance level and the corresponding fourth significance level is greater than a preset second overall significance threshold; if so, the point reaches the second overall significance level.

[0179] The product of the third significance level and the corresponding fourth significance level can be expressed as:

[0180] ;

[0181] Among them, the represents the third level of significance, the represents the fourth level of significance, the represents the product of the third significance level and the corresponding fourth significance level. Can be set to 0.8 (experience value), if If it is greater than 0.8, the point reaches the second overall significance level.

[0182] Fig. 9 This is a flow chart of a process for obtaining a third saliency level provided in a ninth embodiment of the present invention. The process for obtaining a third saliency level includes:

[0183] S901. Obtain a third slope corresponding to the cth point on the third curve graph, a third maximum slope and a third minimum slope corresponding to the points on the third curve graph, wherein c is a positive integer in the range of 1-15.

[0184] The third slope can be expressed as:

[0185] ;

[0186] Among them, the is the number of newly added pixels corresponding to the c+1th point, is the number of newly added pixels corresponding to the c-1th point, represents the preset threshold value, the represents the third slope.

[0187] According to the above process of obtaining the third slope, the slopes corresponding to the various points on the third curve graph can be obtained by analogy, wherein the one with the largest slope is the third maximum slope, and the one with the smallest slope is the third minimum slope.

[0188] S902. Subtract the third minimum slope from the third slope and divide the resultant value by the third maximum slope to obtain the third significance level of the c-th point with respect to the mutation feature.

[0189] The third significance of the c-th point with respect to the mutation feature can be expressed as:

[0190] ;

[0191] Among them, the represents the first slope, the represents the second minimum slope, the represents the second maximum slope, the is the third significance degree of the c-th point with respect to the mutation feature.

[0192] Fig.10This is a flow chart of a process for obtaining a fourth saliency level provided in a tenth embodiment of the present invention. The process for obtaining a fourth saliency level includes:

[0193] S1001. Obtain a fourth slope corresponding to the c-th point on the fourth curve graph, where c is a positive integer ranging from 1 to 21.

[0194] The fourth slope can be expressed as:

[0195] ;

[0196] Among them, the is the average distance corresponding to the c+1th point, is the average distance corresponding to the c-1th point, d represents the preset threshold value, is the fourth slope.

[0197] S1002. Add a preset fifth value to the fourth slope and take the reciprocal value as the fourth significance level of the c-th point with respect to the clustering feature.

[0198] The smaller the average distance between the newly added pixels and the central pixel, the more significant the clustering feature of the newly added pixels in the window. The smaller it is, the smaller the change in the average distance is, which means that the clustering characteristics of the newly added pixels are more significant, so the fourth significance level is inversely proportional to the fourth slope.

[0199] The fourth significance level can be expressed as:

[0200] ;

[0201] Among them, the For the fourth degree of significance, is the fourth slope, the The fifth value is preferably 1 to prevent the denominator from being zero.

[0202] The present invention has the following beneficial effects:

[0203] In the present invention, first, a mixed image of water bloom and fallen leaves is collected, and the mixed image of water bloom and fallen leaves is grayed to obtain a mixed gray image of water bloom and fallen leaves. The mixed gray image of water bloom and fallen leaves is the image basis for subsequent analysis. Secondly, the mixed gray image of water bloom and fallen leaves is threshold segmented according to a preset fallen leaf threshold to distinguish the fallen leaf area and the suspected water bloom area. Here, the fallen leaf area and the suspected water bloom area are first divided, and the suspected water bloom area needs to be further determined whether it is a water bloom area. Furthermore, the suspected water bloom area is threshold segmented according to a preset water bloom threshold to obtain the water bloom area. Here, the water bloom area is determined by the suspected water bloom area. Finally, the ratio of the area of ​​the water bloom area to the area of ​​the mixed image of water bloom and fallen leaves is obtained, and the degree of water bloom pollution and the countermeasures are judged according to the ratio. Here, the degree of water bloom pollution is judged according to the water bloom pollution degree, and corresponding countermeasures are taken according to the water bloom pollution degree, so that the water bloom pollution situation is alleviated or solved. By adopting the present invention, the degree of water bloom pollution can be obtained according to the mixed image of water bloom and fallen leaves, and corresponding measures can be taken in time to deal with these water blooms according to the degree of water bloom pollution, so as to avoid excessive water blooms causing water quality deterioration, aquatic biological diseases and deaths, ecosystem degradation and biodiversity reduction, economic losses and other losses in aquaculture.

[0204] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0205] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for monitoring pollutants in an aquaculture area, characterized in that: The method comprises: Collecting a mixed image of water blooms and fallen leaves, and graying the mixed image of water blooms and fallen leaves to obtain a mixed grayscale image of water blooms and fallen leaves; Performing threshold segmentation on the mixed grayscale image of water bloom and fallen leaves according to a preset fallen leaf threshold to distinguish fallen leaf areas and suspected water bloom areas; Performing threshold segmentation on the suspected algal bloom area according to a preset algal bloom threshold to obtain an algal bloom area; Obtaining a ratio of the area of ​​the water bloom region to the area of ​​the mixed image of water blooms and fallen leaves, and judging the degree of water bloom pollution and countermeasures according to the ratio; The method further comprises: Segmenting the mixed grayscale image of water bloom and fallen leaves from small to large according to a segmentation threshold, to obtain a preset number of segmented images of water bloom and fallen leaves, wherein a first value range of the segmentation threshold is 0 to 225; Denoising the water bloom and fallen leaves segmented image, so that a value range of a segmentation threshold corresponding to the denoised water bloom and fallen leaves segmented image is narrowed from the first value range to a second value range; Performing Otsu threshold segmentation on the mixed grayscale image of water bloom and fallen leaves within the second value range to obtain an optimal segmentation threshold; Dividing a first threshold range and a second threshold range according to the optimal segmentation threshold, wherein within the second value range, the segmentation threshold before the optimal segmentation threshold is the first threshold range, and the segmentation threshold after the optimal segmentation threshold is the second threshold range; The process of obtaining the leaf falling threshold comprises: Performing threshold segmentation on the mixed grayscale image of water bloom and fallen leaves within the first threshold range and extracting a first initial threshold segmentation image corresponding to the pixel point where fallen leaves first appear, wherein the segmentation threshold corresponding to the first initial threshold segmentation image is the first initial segmentation threshold; Locating the fallen leaf pixels in the first initial threshold segmentation image, clustering the fallen leaf pixels, determining the central pixel of each class and establishing a window so that each window contains all the fallen leaf pixels in the corresponding class; The segmentation threshold of the mixed grayscale image of water blooms and fallen leaves is increased from the first initial segmentation threshold by a preset threshold value, and the number of newly added pixels in any of the windows and the average distance between the newly added pixels and the central pixel are obtained; Establishing a first curve graph with the number of newly added pixels as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis; Establishing a second curve graph with the average distance as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis; Obtaining points reaching a first overall significance level on the first curve graph and the second curve graph, and obtaining segmentation thresholds corresponding to the points; Arrange the segmentation thresholds from small to large, and calculate the difference between adjacent segmentation thresholds; Determine whether the value obtained by adding the preset first value to the average value of the difference and then taking the inverse value is greater than the preset first threshold value, and if so, the maximum segmentation threshold value is the leaf falling threshold value; The pixel point area segmented by the largest segmentation threshold is the fallen leaf area, and all non-fallen leaf areas are suspected water bloom areas; Performing threshold segmentation on the suspected algal bloom area according to a preset algal bloom threshold to obtain an algal bloom area includes: Performing threshold segmentation on the suspected algal bloom area within the second threshold range and extracting a second initial threshold segmentation image corresponding to the initially appearing pixel point, wherein the segmentation threshold corresponding to the second initial threshold segmentation image is the second initial segmentation threshold; Locating the pixel points in the second initial threshold segmented image, clustering the pixel points, determining the central pixel point of each class and establishing a window so that each window contains all the pixel points in the corresponding class; The segmentation threshold of the threshold segmented image is increased sequentially by a preset threshold value, and the number of newly added pixel points in any of the windows and the average distance between the newly added pixel points and the central pixel point are obtained; Establishing a third curve graph with the number of newly added pixels as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis; Establishing a fourth curve graph with the average distance as the vertical axis and the number of times the preset threshold value is increased as the horizontal axis; Obtaining points reaching a second overall significance level on the third curve graph and the fourth curve graph, and obtaining segmentation thresholds corresponding to the points; The segmentation threshold is subtracted from a preset third value and the absolute value is obtained, and the number of channels after adding a preset fourth value to the absolute value is obtained to obtain the possibility; The segmentation threshold corresponding to the one with the greatest possibility is the water bloom threshold; the suspected water bloom area is segmented using the water bloom threshold, and the pixel point area greater than or equal to the water bloom threshold is the water bloom area.

2. The method for monitoring pollutants in an aquaculture area according to claim 1, characterized in that: Obtaining points reaching a first overall significance level on the first curve graph and the second curve graph includes: Obtaining a first significance level of each point on the first curve graph with respect to a mutation feature and a second significance level of each point on the second curve graph with respect to an aggregation feature, wherein the first significance level corresponds to the second significance level in one-to-one correspondence; It is determined whether the product of the first significance level and the corresponding second significance level is greater than a preset first overall significance threshold; if so, the point reaches the first overall significance level.

3. The method for monitoring pollutants in an aquaculture area according to claim 2, characterized in that: The process of obtaining the first significance level includes: Obtaining a first slope corresponding to the cth point on the first curve graph, a first maximum slope and a first minimum slope corresponding to the points on the first curve graph, wherein c is a positive integer in the range of 1-21; A value obtained by subtracting the first minimum slope from the first slope and dividing the result by the first maximum slope is the first significance level of the c-th point with respect to the mutation feature.

4. The method for monitoring pollutants in an aquaculture area according to claim 3, characterized in that: The process of obtaining the second significance level includes: Obtaining a second slope corresponding to the c-th point on the second curve graph, where c is a positive integer in the range of 1-21; The reciprocal value of the second slope plus the preset second value is calculated as the second significance level of the c-th point with respect to the clustering feature.

5. The method for monitoring pollutants in an aquaculture area according to claim 4, characterized in that: Obtaining the points reaching the second overall significance level on the third curve graph and the fourth curve graph comprises: Obtaining a third significance level of each point on the third curve graph with respect to the mutation feature and a fourth significance level of each point on the fourth curve graph with respect to the aggregation feature, wherein the third significance level corresponds to the fourth significance level in a one-to-one manner; It is determined whether the product of the third significance level and the corresponding fourth significance level is greater than a preset second overall significance threshold; if so, the point reaches the second overall significance level.

6. The method for monitoring pollutants in an aquaculture area according to claim 5, characterized in that: The process of obtaining the third significance level includes: Obtaining a third slope corresponding to the cth point on the third curve graph, a third maximum slope and a third minimum slope corresponding to the points on the third curve graph, wherein c is a positive integer in the range of 1-15; The value obtained by subtracting the third minimum slope from the third slope and dividing the result by the third maximum slope is the third significance level of the c-th point with respect to the mutation feature.

7. The method for monitoring pollutants in an aquaculture area according to claim 1, characterized in that: According to the ratio, the degree of algal bloom pollution and the corresponding measures include: When the ratio is less than the first pollution threshold, the water bloom pollution level is light, and a first response measure is taken, wherein the first response measure is to add a microbial agent to the water bloom pollution area to prevent the growth of water blooms; When the ratio is greater than or equal to the first pollution threshold and less than or equal to the second pollution threshold, the degree of water bloom pollution is moderate, and a second response measure is taken, wherein the second response measure is to set up a fence and add an aerator to the water bloom pollution area; When the ratio is greater than the second pollution threshold, the water bloom pollution level is severe, and a third response measure is taken, which is to spray flocculants on the water bloom pollution area and add a preset amount of activated carbon adsorbent.

Images