Garbage processing data analysis optimization method and system under cloud computing platform

Abstract

The invention relates to the technical field of garbage identification, in particular to a garbage processing data analysis optimization method and system under a cloud computing platform. The method comprises the following steps: acquiring chromaticity similarity coefficients of suspected junk areas among different stable frame images; in combination with the position features of the suspected junk areas between the different stable frame images, obtaining multiple groups of similar suspected junk areas between the different stable frame images; obtaining the floating passive degree of each group of similar garbage suspected areas between different adjacent stable frame images, and screening out garbage determined area groups; according to the morphological features, texture features and corresponding floating passivity degrees of different garbage determination areas in each group of garbage determination areas, obtaining garbage matching priority of each garbage determination area in each group of garbage determination areas; and garbage to-be-matched areas in each group of garbage determination areas are screened out, and garbage is identified. The garbage recognition efficiency is improved by obtaining the accurate matching area of the garbage.

Description

Technical Field

[0001] The present invention relates to the technical field of garbage identification, and in particular to a garbage disposal data analysis and optimization method and system under a cloud computing platform. Background Art

[0002] In the marine plastic waste recycling project, in order to improve the recycling efficiency, we first use an underwater robot equipped with a camera to conduct preliminary reconnaissance of the amount and distribution of plastic waste in the seabed area, and then use the cloud computing platform to identify the garbage based on the reconnaissance images and formulate a garbage recycling plan. Therefore, it is necessary to accurately identify the garbage to improve the garbage recycling efficiency.

[0003] In the prior art, garbage identification and classification affected by seabed garbage reconnaissance is achieved through template matching algorithms. However, due to the interference of swimming fish in the ocean, there are many objects to be identified by template matching, which consumes a lot of redundant platform cloud computing resources. In addition, the seabed environment is complex, which leads to serious distortion of garbage areas in some frame images, affecting the template matching accuracy and poor recognition efficiency of garbage images. Summary of the invention

[0004] In order to solve the technical problems of serious distortion in garbage areas, affecting template matching accuracy and poor image recognition efficiency, the purpose of the present invention is to provide a garbage disposal data analysis optimization method and system under a cloud computing platform. The technical solutions adopted are as follows:

[0005] The present invention proposes a garbage disposal data analysis and optimization method under a cloud computing platform, the method comprising:

[0006] Obtain video frame images of low-flow sea areas containing garbage at every moment, as well as water flow velocity data;

[0007] According to the grayscale distribution of pixels between different video frame images, the stable frame images are screened out to obtain multiple suspected garbage areas in each stable frame image; according to the color difference of the suspected garbage areas between different stable frame images, the chromaticity similarity coefficient of the suspected garbage areas between different stable frame images is obtained;

[0008] According to the position characteristics of the suspected garbage areas between different stable frame images and the chromaticity similarity coefficient, multiple groups of similar suspected garbage areas between different stable frame images are obtained; according to the difference change characteristics of each group of similar suspected garbage areas between different adjacent stable frame images and the water flow rate data distribution of the stable frame images, the floating passivity of each group of similar suspected garbage areas between different adjacent stable frame images is obtained, and the garbage determination area group is screened out;

[0009] According to the morphological features, texture features and corresponding floating passivity of different garbage determination areas in each group of garbage determination areas, the garbage matching priority of each garbage determination area in each group of garbage determination areas is obtained; according to the garbage matching priority of each garbage determination area in different groups of garbage determination areas, the garbage to-be-matched areas in each group of garbage determination areas are screened out;

[0010] Identify garbage based on the area where it is to be matched.

[0011] Furthermore, the image stability of each video frame image is obtained according to the grayscale distribution of pixels between different video frame images;

[0012] If the image stability degree of each video frame image is greater than a preset stability threshold, the corresponding video frame image is used as a stable frame image.

[0013] Furthermore, the method for obtaining the image stability comprises:

[0014] For any video frame image, count the number of pixels corresponding to each gray level, and construct a gray level performance curve with the gray level as the horizontal axis and the number of corresponding pixels as the vertical axis;

[0015] Obtaining the average value of the correlation coefficient of the grayscale performance curve between each video frame image and different other video frame images as the grayscale correlation of each video frame image relative to other video frame images;

[0016] The maximum value of the corresponding grayscale correlation in all video frame images is obtained as the grayscale reference correlation; the ratio of the corresponding grayscale correlation of each video frame image to the grayscale reference correlation is obtained as the image stability of each video frame image.

[0017] Furthermore, the method for obtaining the suspected garbage area includes:

[0018] K-means clustering is performed on all pixels according to the grayscale values ​​of the pixels in each stable frame image to obtain multiple pixel clustering areas as suspected garbage areas.

[0019] Furthermore, the method for obtaining the chromaticity similarity coefficient includes:

[0020] Obtain the parameter value of each pixel point in each suspected garbage area in each stable frame image in the LAB space;

[0021] For any suspected garbage area in any stable frame image as the target area, the number of pixels with the same corresponding parameter values ​​between different suspected garbage areas and the target area in each other stable frame image is counted as the number of overlapping pixels;

[0022] The overlapping pixel data between each suspected garbage area and the target area in each other stable frame image and the ratio of the maximum number of overlapping pixels are obtained as the chromaticity similarity coefficient between each suspected garbage area and the target area in each other stable frame image; the target area is changed to obtain the chromaticity similarity coefficient of the suspected garbage area between different stable frame images.

[0023] Furthermore, the method for obtaining the suspected area of ​​the same type of garbage includes:

[0024] Obtaining the geometric center of each suspected garbage area on each stable frame image; obtaining the relative distance between the corresponding geometric centers of the suspected garbage areas between different stable frame images as the area distance;

[0025] Perform negative correlation mapping on the regional distance, calculate the product of the negative correlation mapping result and the chromaticity similarity coefficient of the garbage suspected area between the corresponding stable frame images, and normalize them as the regional similarity of the garbage suspected area between different stable frame images;

[0026] If the regional similarity of the suspected garbage areas between different stable frame images is greater than a preset similarity threshold, the corresponding suspected garbage areas between the different stable frame images are taken as similar suspected garbage areas, and similar suspected garbage areas of each suspected garbage area are obtained to form multiple groups of similar suspected garbage areas.

[0027] Furthermore, the step of obtaining the floating passivity of each group of similar garbage suspected areas between different adjacent stable frame images and screening out garbage confirmed area groups includes:

[0028] The number of different pixels of each group of suspected garbage areas of the same type between adjacent stable frame images is obtained as the floating intensity parameter of the suspected garbage areas of the same type between adjacent stable frame images;

[0029] Obtain the DTW distance between the curve composed of the floating intensity parameters and the curve composed of the water flow rate data of each group of similar garbage suspected areas at the corresponding moment, and perform negative correlation normalization mapping on the DTW distance as the floating passivity of each group of similar garbage suspected areas between different adjacent stable frame images;

[0030] If the floating passivity of each group of similar garbage suspected areas is greater than a preset threshold, the corresponding group of similar garbage suspected areas will be used as garbage confirmed areas.

[0031] Furthermore, the method for obtaining the garbage matching priority includes:

[0032] For each group of garbage determination areas, the number of pixels in each garbage determination area is obtained as a morphological feature; the morphological feature of each garbage determination area is normalized, and the product of the normalized result and the corresponding floating passive intensity is calculated as the garbage target degree of each garbage determination area in each group of garbage determination areas;

[0033] Obtain the corner points of each garbage determination area based on the corner point detection algorithm; obtain the gradient mean of all pixel points in the neighborhood of each corner point as the texture saliency value of each corner point; obtain the texture saliency mean of all corner points in each garbage determination area as the texture performance level of each determination area;

[0034] The product of the garbage target degree and the texture expression level of each garbage determination area in each group of garbage determination areas is obtained and normalized and mapped as the garbage matching priority of each area in each group of garbage determination areas.

[0035] Furthermore, the method for obtaining the garbage to-be-matched area includes:

[0036] The garbage matching priority value of all garbage determination areas in each group of garbage determination areas is selected to be the largest, and the corresponding garbage determination area is used as the garbage to-be-matched area.

[0037] The present invention also proposes a garbage disposal data analysis and optimization system under a cloud computing platform, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of the garbage disposal data analysis and optimization method under the cloud computing platform.

[0038] The present invention has the following beneficial effects:

[0039] The present invention selects stable frame images according to the grayscale distribution of pixel points between different video frame images, and can select frames with stable image quality for analysis, thereby reducing the complexity and error of subsequent processing, and obtaining multiple suspected garbage areas in each stable frame image, which is helpful to narrow the processing scope and improve recognition efficiency; according to the color difference of the suspected garbage areas between different stable frame images, the chromaticity similarity coefficient of the suspected garbage areas between different stable frame images is obtained, which reflects the color consistency of garbage between different frames and is helpful to identify objects with similar color features; according to the position characteristics of the suspected garbage areas between different stable frame images and the chromaticity similarity coefficient, multiple groups of similar suspected garbage areas between different stable frame images are obtained, which is helpful to reduce redundant information and improve recognition accuracy; according to the position characteristics of each group of similar garbage areas between different adjacent stable frame images, the chromaticity similarity coefficient of each group of similar garbage areas between different adjacent stable frame images is obtained, which is helpful to reduce redundant information and improve recognition accuracy. The difference change characteristics of the suspected garbage area and the water flow rate data distribution of the stable frame image are obtained to obtain the floating passivity of each group of similar suspected garbage areas between different adjacent stable frame images, and the garbage determination area group is screened out, which reflects the stability and dynamic change characteristics of the garbage in the video frame image, and the garbage determination area group with high stability and small dynamic change is screened out; according to the morphological characteristics, texture characteristics and corresponding floating passivity of different garbage determination areas in each group of garbage determination areas, the garbage matching priority of each garbage determination area in each group of garbage determination areas is obtained, and the garbage to-be-matched area in each group of garbage determination areas is screened out, and the characteristics of each garbage determination area are more comprehensively evaluated, so as to more accurately identify the type and attributes of the garbage, and the garbage to-be-matched area with high matching priority can be screened out; the garbage is identified. The present invention improves the garbage identification efficiency by obtaining the accurate matching area of ​​the garbage. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] 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.

[0041] Figure 1 A flowchart of a method for analyzing and optimizing garbage disposal data on a cloud computing platform provided by an embodiment of the present invention;

[0042] Figure 2 A flow chart of a method for obtaining suspected areas of similar garbage provided by an embodiment of the present invention;

[0043] Figure 3 A flow chart of a method for obtaining garbage matching priority provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0044] 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 garbage disposal data analysis optimization method and system under a cloud computing platform proposed by the present invention, its specific implementation method, 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.

[0045] 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.

[0046] The following is a detailed description of a garbage disposal data analysis and optimization method and system under a cloud computing platform provided by the present invention in conjunction with the accompanying drawings.

[0047] See also Figure 1 , which shows a method flow chart of a garbage disposal data analysis and optimization method under a cloud computing platform provided by an embodiment of the present invention, specifically comprising:

[0048] Step S1: Obtain video frame images of low-flow sea areas containing garbage at each moment, as well as water flow rate data.

[0049] In the embodiment of the present invention, in order to avoid pollution and damage to the marine environment and improve the efficiency of garbage recycling, it is necessary to accurately identify garbage; first, an underwater robot with a camera is used to collect video in the low-velocity sea area, and the video frame image is subsequently analyzed; the water velocity data of the shooting sea area is collected by the water velocity sensor configured by the underwater robot, and uploaded to the cloud acquisition system for subsequent processing. The video frame image of the low-velocity sea area containing garbage at each moment and the water velocity data are obtained.

[0050] It should be noted that, in an embodiment of the present invention, the acquired video frame images are placed in the same mapping image according to coordinates, that is, the left vertical edge in the video frame image is used as the longitudinal axis of the coordinate system, and the lower edge perpendicular to the longitudinal axis is used as the horizontal axis of the coordinate system. The position of each pixel point on each image is coordinateized to facilitate subsequent analysis of the image.

[0051] It should be noted that, in one embodiment of the present invention, video images collected within the history of 10s at the real-time moment are obtained with a time interval of 0.2s, that is, 0.2s is taken as one frame, and the video images are converted into static frame images to obtain multiple video frame images; in other embodiments of the present invention, the time range and time interval can be set according to the specific situation, which is not limited or elaborated here.

[0052] Step S2: According to the grayscale distribution of pixels between different video frame images, the stable frame images are screened out to obtain multiple suspected garbage areas in each stable frame image; according to the color difference of the suspected garbage areas between different stable frame images, the chromaticity similarity coefficient of the suspected garbage areas between different stable frame images is obtained.

[0053] Considering that fish close to the camera will cover a large area of ​​garbage, and tiny swimming will cause large differences in imaging performance, resulting in changes in the grayscale values ​​of pixels in the image, by analyzing the grayscale distribution of pixels in different video frames, it can better reflect the changes in image quality and the coverage of garbage areas. Screening out stable frame images can help analyze the specific form of garbage; according to the grayscale distribution of pixels between different video frames, stable frame images are screened out to obtain multiple suspected garbage areas in each stable frame image.

[0054] Preferably, in one embodiment of the present invention, the method for acquiring a stable frame image includes:

[0055] According to the grayscale distribution of pixels between different video frame images, the image stability of each video frame image is obtained;

[0056] In one embodiment of the present invention, a method for obtaining image stability includes:

[0057] For any video frame image, count the number of pixels corresponding to each gray level, and construct a gray level performance curve with the gray level as the horizontal axis and the number of corresponding pixels as the vertical axis;

[0058] Obtaining the average value of the correlation coefficient of the grayscale performance curve between each video frame image and different other video frame images as the grayscale correlation of each video frame image relative to other video frame images;

[0059] The maximum value of the corresponding grayscale correlation in all video frame images is obtained as the grayscale reference correlation; the ratio of the corresponding grayscale correlation of each video frame image to the grayscale reference correlation is obtained as the image stability of each video frame image.

[0060] It should be noted that, in one embodiment of the present invention, the correlation coefficient is the Pearson correlation coefficient; the specific means are technical means well known to those skilled in the art and will not be elaborated here.

[0061] If the image stability degree of each video frame image is greater than a preset stability threshold, the corresponding video frame image is used as a stable frame image.

[0062] It should be noted that, in one embodiment of the present invention, the preset stability threshold is 0.78; in other embodiments of the present invention, the preset stability threshold may be specifically set according to specific circumstances, which is not limited or elaborated herein.

[0063] In an image, different objects or regions will present different grayscale values ​​due to factors such as their material, color, and lighting conditions. Through K-means clustering, the pixels in the image can be divided into different clustering areas. The clustering areas may correspond to different objects or areas and be analyzed as suspected garbage areas.

[0064] Preferably, in one embodiment of the present invention, the method for obtaining the suspected garbage area includes:

[0065] K-means clustering is performed on all pixels according to the grayscale values ​​of the pixels in each stable frame image to obtain multiple pixel clustering areas as suspected garbage areas.

[0066] The K-means clustering algorithm clusters multiple cluster objects into K specified clusters based on the similarity between the cluster objects. Each cluster object belongs to and only belongs to one cluster with the smallest center distance and grayscale distance to the suspected garbage area. It should be noted that in one embodiment of the present invention, when the K-means clustering algorithm is used to cluster all pixels, the method for obtaining the K value is: the elbow rule is used to determine the K value, and the corresponding number of pixel cluster areas are obtained as suspected garbage areas. The specific K-means clustering algorithm and the elbow rule are technical means well known to those skilled in the art and will not be elaborated here.

[0067] The same seabed object should have similar color performance in different frame images. By analyzing the color differences of suspected garbage areas between different stable frame images, its color uniformity and difference can be evaluated, and the color similarity of suspected garbage areas in different stable frame images can be quantified, which helps to identify garbage more accurately. According to the color differences of suspected garbage areas between different stable frame images, the chromaticity similarity coefficient of suspected garbage areas between different stable frame images is obtained.

[0068] Preferably, in one embodiment of the present invention, the method for obtaining the chromaticity similarity coefficient includes:

[0069] LAB space is the color space with the widest color gamut and can describe and compare colors more accurately. The parameter values ​​of the pixels of each suspected garbage area in each stable frame image in LAB space are obtained. For any suspected garbage area in any stable frame image as the target area, the number of pixels with the same corresponding parameter values ​​between different suspected garbage areas and the target area in each other stable frame image is counted as the number of overlapping pixels.

[0070] The overlapping pixel data between each suspected garbage area and the target area in each other stable frame image and the ratio of the maximum number of overlapping pixels are obtained as the chromaticity similarity coefficient between each suspected garbage area and the target area in each other stable frame image; the target area is changed to obtain the chromaticity similarity coefficient of the suspected garbage area between different stable frame images.

[0071] Based on this, obtaining the chromaticity similarity coefficient of any two suspected garbage areas between different stable frame images for analysis is more helpful in evaluating the possibility that the two suspected garbage areas are the same seabed object.

[0072] Step S3: According to the position characteristics of the suspected garbage areas between different stable frame images and the chromaticity similarity coefficient, multiple groups of similar suspected garbage areas between different stable frame images are obtained; according to the difference change characteristics of each group of similar suspected garbage areas between different adjacent stable frame images and the water flow rate data distribution of the stable frame images, the floating passivity of each group of similar suspected garbage areas between different adjacent stable frame images is obtained, and the garbage determination area group is screened out.

[0073] The position difference of the same object in adjacent frame images changes little. The chromaticity similarity coefficient can quantify the color similarity between different suspected garbage areas. When the chromaticity similarity coefficient of two areas is high, it indicates that the color features between the areas are similar and may belong to the same object. Therefore, by combining the position features and the chromaticity similarity coefficient, similar suspected garbage areas in different stable frame images can be identified more accurately. According to the position features of the suspected garbage areas between different stable frame images and the chromaticity similarity coefficient, multiple groups of similar suspected garbage areas between different stable frame images are obtained.

[0074] Preferably, in one embodiment of the present invention, the method for obtaining the suspected area of ​​the same type of garbage can refer to Figure 2 , which shows a flow chart of a method for obtaining suspected areas of similar garbage, including:

[0075] Step S201: obtaining the geometric center of each suspected garbage region on each stable frame image; obtaining the relative distance between the corresponding geometric centers of the suspected garbage regions between different stable frame images as the region distance.

[0076] In order to better analyze the matching relationship of suspected garbage areas in stable frame images, the distance between the geometric centers of suspected garbage areas is analyzed, which helps to quantify the position changes between areas. The larger the relative distance and the larger the area distance, the smaller the possibility of the same object.

[0077] It should be noted that, in the embodiments of the present invention, the geometric center of each suspected garbage area can be obtained by a shape boundary centroid algorithm; or the geometric center can be obtained by calculating the coordinate mean of all pixel points in the suspected garbage area; the specific means are technical means well known to those skilled in the art and will not be elaborated here.

[0078] Step S202: negative correlation mapping is performed on the regional distance, and the product of the negative correlation mapping result and the chromaticity similarity coefficient of the garbage suspected area between the corresponding stable frame images is calculated, and normalized as the regional similarity of the garbage suspected area between different stable frame images.

[0079] The larger the regional distance, the greater the change in position difference, the smaller the possibility that the two regions are the same object, and the smaller the regional similarity, which is a negative correlation.

[0080] In one embodiment of the present invention, the formula for region similarity is expressed as:

[0081]

[0082] Among them, S x,y,u,r V represents the regional similarity between the yth suspected garbage region in the xth stable frame image and the rth suspected garbage region in the uth stable frame image; x,y,u,r represents the chromaticity similarity coefficient between the yth suspected garbage region in the xth stable frame image and the rth suspected garbage region in the uth stable frame image; D x,y,u,r It represents the relative distance between the corresponding geometric centers of the yth suspected garbage area in the xth stable frame image and the rth suspected garbage area in the uth stable frame image, that is, the area distance; norm() represents the normalization function.

[0083] In the formula of regional similarity, D x,y,u,r In order to avoid the denominator being 0, 0.01 is added to +0.01, and the formula is meaningless; the larger the regional distance between suspected garbage areas, the larger the position difference, and the smaller the credibility of regional similarity; the smaller the regional distance between suspected garbage areas, the smaller the position difference, the greater the credibility of regional similarity, and the larger the chromatic similarity coefficient, the greater the possibility that they are the same object, and the greater the regional similarity.

[0084] It should be noted that, in some embodiments of the present invention, existing distance algorithms such as Euclidean distance and Manhattan distance may be used to obtain the relative distance between geometric centers. The specific means are well known to those skilled in the art and will not be elaborated here.

[0085] Step S203: If the regional similarity of the suspected garbage regions between different stable frame images is greater than a preset similarity threshold, the corresponding suspected garbage regions between the different stable frame images are taken as similar suspected garbage regions, and similar suspected garbage regions of each suspected garbage region are obtained to form multiple groups of similar suspected garbage regions.

[0086] It should be noted that, in one embodiment of the present invention, the preset similarity threshold is 0.88; in other embodiments of the present invention, the preset similarity threshold may be set according to specific circumstances, which will not be limited or elaborated herein.

[0087] The plastic garbage on the seabed floats locally with the flow of seawater. The floating pattern of garbage should be positively correlated with the flow rate of seawater. However, fish have independent life consciousness compared to garbage, which does not conform to the relevant laws of seawater fluidity. Therefore, by analyzing the difference change characteristics of each group of suspected garbage areas of the same type, the position change of objects in the area can be reflected. Combined with the distribution of water flow rate data, the degree of passive floating of the suspected garbage areas can be quantitatively evaluated, and then the garbage determination area group can be screened out, which is helpful for analyzing the characteristics of the garbage. According to the difference change characteristics of each group of suspected garbage areas of the same type between different adjacent stable frame images, and the water flow rate data distribution of the stable frame images, the floating passive degree of each group of suspected garbage areas of the same type between different adjacent stable frame images is obtained, and the garbage determination area group is screened out.

[0088] Preferably, in one embodiment of the present invention, obtaining the floating passivity of each group of similar garbage suspected areas between different adjacent stable frame images and screening out garbage confirmed area groups includes:

[0089] The number of different pixels of each group of suspected garbage areas of the same type between adjacent stable frame images is obtained as the floating intensity parameter of each group of suspected garbage areas of the same type between adjacent stable frame images;

[0090] The DTW distance between the curve composed of the floating intensity parameters and the curve composed of the water flow rate data at the corresponding moment of each group of suspected garbage areas of the same type is obtained, and the DTW distance is negatively correlated and normalized to be used as the floating passivity of each group of suspected garbage areas of the same type between different adjacent stable frame images;

[0091] In one embodiment of the present invention, the formula for floating passivity is expressed as:

[0092]

[0093] Among them, IB h Indicates the floating passivity of the h-th group of suspected garbage areas; DTW hIt represents the DTW distance between the curve composed of the floating intensity parameters and the curve composed of the water flow rate data at the corresponding moment of the h-th group of suspected garbage areas of the same type; norm() represents the normalization function.

[0094] In the formula for floating passivity, DTW h +0.01 is added to avoid the denominator being 0, so the formula is meaningless; the curve composed of the floating intensity parameters and water flow rate data of each group of similar garbage suspected areas at the corresponding moment in the hth group, the larger the DTW distance between the composed curves, the smaller the similarity between the time series changes of the object's floating and the water flow rate, the stronger the object's sense of autonomous action, the less passive the floating, and the smaller the possibility of being garbage.

[0095] If the floating passivity of each group of similar garbage suspected areas is greater than a preset threshold, the corresponding group of similar garbage suspected areas will be used as garbage confirmed areas.

[0096] It should be noted that, in one embodiment of the present invention, the size of the preset degree threshold is 0.8; in other embodiments of the present invention, the size of the preset degree threshold can be set according to specific circumstances, and is not limited or elaborated herein.

[0097] Step S4: Obtain the garbage matching priority of each garbage determination area in each group of garbage determination areas according to the morphological features, texture features and corresponding floating passivity of different garbage determination areas in each group of garbage determination areas; and screen out the garbage to-be-matched areas in each group of garbage determination areas according to the garbage matching priority of each garbage determination area in different groups of garbage determination areas.

[0098] Due to the low direct coverage rate between the garbage and the camera at certain moments, fewer feature points are exposed, which increases the difficulty of subsequent garbage template matching. Therefore, the morphological characteristics of the garbage determination area are analyzed. The larger the morphological characteristics, the greater the priority of garbage matching; there are many biological metabolites inside the seawater, which leads to complex changes in seawater turbidity and distortion of some garbage areas. Therefore, the texture characteristics of the garbage determination area are analyzed. The more obvious the texture characteristics, the more obvious the displayed characteristics, and the more conducive to garbage identification; the floating passivity of the garbage determination area indicates the degree of correlation between the objects in the area and the flow of seawater. The greater the floating passivity, the more likely it is that the floating is caused by the flow of seawater, and the greater the correlation, the more likely it is a garbage area; therefore, according to the morphological characteristics, texture characteristics and corresponding floating passivity of different garbage determination areas in each group of garbage determination areas, the garbage matching priority of each garbage determination area in each group of garbage determination areas is obtained.

[0099] Preferably, in one embodiment of the present invention, the method for obtaining the garbage matching priority is as follows: Figure 3, which shows a flow chart of a method for obtaining garbage matching priority, including:

[0100] Step S301: For each group of garbage determination areas, obtain the number of pixels in each garbage determination area as a morphological feature; normalize the morphological feature of each garbage determination area, and calculate the product of the normalized result and the corresponding floating passive intensity as the garbage target degree of each garbage determination area in each group of garbage determination areas.

[0101] The number of pixels as a morphological feature can intuitively reflect the size and shape of the garbage area. The larger the morphological feature, the more feature points it shows. Normalization processing can map the feature values ​​of different areas to the same scale, which is convenient for subsequent comparison and analysis. The floating passive intensity is used as the confidence level for judging whether it is garbage. The larger the floating passive intensity, the more likely it is to conform to the relevant laws of seawater fluidity, and the more comprehensive and accurate the assessment of the garbage target degree.

[0102] In one embodiment of the present invention, the formula for the garbage target level is expressed as:

[0103]

[0104] Among them, IE m,n IB represents the garbage target level of the nth garbage determination area in the mth group of garbage determination areas; m Indicates the floating passivity of the mth group of garbage determination area; Q m,n represents the number of pixels in the nth garbage determination area of ​​the mth group of garbage determination areas, that is, the morphological characteristics of the nth garbage determination area; Q m,max It represents the maximum number of pixels in all garbage determination areas in the mth group of garbage determination areas.

[0105] In the formula for the garbage target level, It represents the ratio of the number of pixels in the nth garbage determination area in the mth group of garbage determination areas to the maximum number of pixels in all garbage determination areas, that is, the morphological features of each garbage determination area are normalized. The larger the ratio, the larger the normalization result. The larger the number of pixels in the nth garbage determination area in the mth group of garbage determination areas, the larger the morphological features are relatively, and the more likely it is to show more feature information. The greater the floating passivity, the smaller the awareness of autonomous action, and the greater the garbage target degree.

[0106] Step S302: Obtain the corner points of each garbage determination area based on the corner point detection algorithm; obtain the gradient mean of all pixel points in the neighborhood of each corner point as the texture saliency value of each corner point; obtain the texture saliency value mean of all corner points in each garbage determination area as the texture representation level of each garbage determination area.

[0107] In the garbage identification area, corner points may correspond to the edges, inflection points or key structural points of the garbage. The corner detection algorithm can accurately identify the corner points in the image. The gradient mean can reflect the degree of texture change in the local area of ​​the image. The gradient mean is calculated in the neighborhood of the corner point to obtain the texture information around the corner point. The texture performance level of each garbage identification area is then evaluated by the mean, which can reflect the overall texture characteristics of the garbage area and help to distinguish different types of garbage.

[0108] It should be noted that, in one embodiment of the present invention, the neighborhood range of a corner point is a range consisting of the corner point pixel as the center and all adjacent pixels; in other embodiments of the present invention, the neighborhood range of a corner point can be set according to specific circumstances and is not limited or elaborated here.

[0109] Step S303: obtaining the product of the garbage target degree and the texture expression level of each garbage determination area in each group of garbage determination areas, and performing normalized mapping as the garbage matching priority of each area in each group of garbage determination areas.

[0110] In one embodiment of the present invention, the formula for garbage matching priority is expressed as:

[0111] IF m,n =norm(IE m,n ×U m,n );

[0112] Among them, IF m,n Indicates the garbage matching priority of the nth garbage determination area in the mth group of garbage determination areas; IE m,n represents the garbage target level of the nth garbage determination area in the mth group of garbage determination areas; U m,n It represents the texture expression level of the nth garbage determination area in the mth group of garbage determination areas; norm() represents the normalization function.

[0113] In the formula of garbage matching priority, the greater the garbage target degree, the more garbage features the area exhibits, the greater the texture expression level, the lower the distortion degree, and the higher the garbage matching priority.

[0114] The garbage matching priority is used to evaluate the possibility of garbage and the performance of texture distortion. The larger the garbage matching limit, the higher the classification accuracy and the richer the texture information. The more information showing the characteristics of garbage, the more conducive to garbage identification. According to the garbage matching priority of each garbage determination area in different groups of garbage determination areas, the garbage areas to be matched in each group of garbage determination areas are screened out.

[0115] Preferably, in one embodiment of the present invention, the method for obtaining the garbage to-be-matched area includes:

[0116] The garbage matching priority value of all garbage determination areas in each group of garbage determination areas is selected to be the largest, and the corresponding garbage determination area is used as the garbage to-be-matched area.

[0117] Step S5: Identify the garbage according to the garbage matching area.

[0118] By accurately identifying the matching areas, the system can more accurately determine the type of garbage, thereby reducing the possibility of misclassification.

[0119] It should be noted that in another embodiment of the present invention, after obtaining the garbage to-be-matched area, the garbage can be identified, including: mapping each garbage to-be-matched area to an image with a grayscale value of 0 to obtain the corresponding garbage to-be-matched image, and the image analysis module in the cloud computing platform performs a template matching algorithm on the garbage to-be-matched image to achieve garbage identification and classification, and then sends the seabed garbage identification and classification results to the seabed garbage recycling center via the data transmission module, which helps to formulate a recycling strategy based on the identification and classification situation, and clean and recycle the seabed plastic garbage according to the recycling strategy through the garbage recovery submersible.

[0120] In summary, the present invention analyzes the grayscale distribution of pixels between different video frame images to obtain the chromatic similarity coefficient of the suspected garbage areas between different stable frame images; combines the positional characteristics of the suspected garbage areas between different stable frame images to obtain multiple groups of similar suspected garbage areas between different stable frame images; obtains the floating passivity of each group of similar suspected garbage areas between different adjacent stable frame images to filter out the garbage determination area group; obtains the garbage matching priority of each garbage determination area in each group of garbage determination areas according to the morphological characteristics, texture characteristics and corresponding floating passivity of different garbage determination areas in each group of garbage determination areas; filters out the garbage to-be-matched areas in each group of garbage determination areas to identify the garbage. The present invention improves the garbage identification efficiency by obtaining accurate garbage matching areas.

[0121] The present invention also proposes a garbage disposal data analysis and optimization system under a cloud computing platform, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of a garbage disposal data analysis and optimization method under a cloud computing platform.

[0122] 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.

[0123] 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 garbage disposal data analysis and optimization method under a cloud computing platform, characterized in that: The method comprises: Obtain video frame images of low-flow sea areas containing garbage at every moment, as well as water flow velocity data; According to the grayscale distribution of pixels between different video frame images, the stable frame images are screened out to obtain multiple suspected garbage areas in each stable frame image; according to the color difference of the suspected garbage areas between different stable frame images, the chromaticity similarity coefficient of the suspected garbage areas between different stable frame images is obtained; According to the position characteristics of the suspected garbage areas between different stable frame images and the chromaticity similarity coefficient, multiple groups of similar suspected garbage areas between different stable frame images are obtained; according to the difference change characteristics of each group of similar suspected garbage areas between different adjacent stable frame images and the water flow rate data distribution of the stable frame images, the floating passivity of each group of similar suspected garbage areas between different adjacent stable frame images is obtained, and the garbage determination area group is screened out; According to the morphological features, texture features and corresponding floating passivity of different garbage determination areas in each group of garbage determination areas, the garbage matching priority of each garbage determination area in each group of garbage determination areas is obtained; according to the garbage matching priority of each garbage determination area in different groups of garbage determination areas, the garbage to-be-matched areas in each group of garbage determination areas are screened out; Identify garbage based on the area where it is to be matched.

2. The method for analyzing and optimizing garbage disposal data under a cloud computing platform according to claim 1, characterized in that: The method for acquiring the stable frame image comprises: According to the grayscale distribution of pixels between different video frame images, the image stability of each video frame image is obtained; If the image stability degree of each video frame image is greater than a preset stability threshold, the corresponding video frame image is used as a stable frame image.

3. The garbage disposal data analysis and optimization method under a cloud computing platform according to claim 2 is characterized in that: The method for obtaining the image stability comprises: For any video frame image, count the number of pixels corresponding to each gray level, and construct a gray level performance curve with the gray level as the horizontal axis and the number of corresponding pixels as the vertical axis; Obtaining the average value of the correlation coefficient of the grayscale performance curve between each video frame image and different other video frame images as the grayscale correlation of each video frame image relative to other video frame images; The maximum value of the corresponding grayscale correlation in all video frame images is obtained as the grayscale reference correlation; the ratio of the corresponding grayscale correlation of each video frame image to the grayscale reference correlation is obtained as the image stability of each video frame image.

4. The garbage disposal data analysis and optimization method under a cloud computing platform according to claim 1 is characterized in that: The method for obtaining the suspected garbage area includes: K-means clustering is performed on all pixels according to the grayscale values ​​of the pixels in each stable frame image to obtain multiple pixel clustering areas as suspected garbage areas.

5. The garbage disposal data analysis and optimization method under a cloud computing platform according to claim 1 is characterized in that: The method for obtaining the chromaticity similarity coefficient includes: Obtain the parameter value of each pixel point in each suspected garbage area in each stable frame image in the LAB space; For any suspected garbage area in any stable frame image as the target area, the number of pixels with the same corresponding parameter values ​​between different suspected garbage areas and the target area in each other stable frame image is counted as the number of overlapping pixels; The overlapping pixel data between each suspected garbage area and the target area in each other stable frame image and the ratio of the maximum number of overlapping pixels are obtained as the chromaticity similarity coefficient between each suspected garbage area and the target area in each other stable frame image; the target area is changed to obtain the chromaticity similarity coefficient of the suspected garbage area between different stable frame images.

6. The garbage disposal data analysis and optimization method under a cloud computing platform according to claim 1 is characterized in that: The method for obtaining the suspected area of ​​the same type of garbage includes: Obtaining the geometric center of each suspected garbage area on each stable frame image; obtaining the relative distance between the corresponding geometric centers of the suspected garbage areas between different stable frame images as the area distance; Perform negative correlation mapping on the regional distance, calculate the product of the negative correlation mapping result and the chromaticity similarity coefficient of the garbage suspected area between the corresponding stable frame images, and normalize them as the regional similarity of the garbage suspected area between different stable frame images; If the regional similarity of the suspected garbage areas between different stable frame images is greater than a preset similarity threshold, the corresponding suspected garbage areas between the different stable frame images are taken as similar suspected garbage areas, and similar suspected garbage areas of each suspected garbage area are obtained to form multiple groups of similar suspected garbage areas.

7. The method for analyzing and optimizing garbage disposal data under a cloud computing platform according to claim 1, characterized in that: The step of obtaining the floating passivity of each group of similar garbage suspected areas between different adjacent stable frame images and screening out garbage confirmed area groups includes: The number of different pixels of each group of suspected garbage areas of the same type between adjacent stable frame images is obtained as the floating intensity parameter of the suspected garbage areas of the same type between adjacent stable frame images; Obtain the DTW distance between the curve composed of the floating intensity parameters and the curve composed of the water flow rate data of each group of similar garbage suspected areas at the corresponding moment, and perform negative correlation normalization mapping on the DTW distance as the floating passivity of each group of similar garbage suspected areas between different adjacent stable frame images; If the floating passivity of each group of similar garbage suspected areas is greater than a preset threshold, the corresponding group of similar garbage suspected areas will be used as garbage confirmed areas.

8. The method for analyzing and optimizing garbage disposal data under a cloud computing platform according to claim 1, characterized in that: The method for obtaining the garbage matching priority includes: For each group of garbage determination areas, the number of pixels in each garbage determination area is obtained as a morphological feature; the morphological feature of each garbage determination area is normalized, and the product of the normalized result and the corresponding floating passive intensity is calculated as the garbage target degree of each garbage determination area in each group of garbage determination areas; Obtain the corner points of each garbage determination area based on the corner point detection algorithm; obtain the gradient mean of all pixel points in the neighborhood of each corner point as the texture saliency value of each corner point; obtain the texture saliency mean of all corner points in each garbage determination area as the texture performance level of each determination area; The product of the garbage target degree and the texture expression level of each garbage determination area in each group of garbage determination areas is obtained and normalized and mapped as the garbage matching priority of each area in each group of garbage determination areas.

9. The garbage disposal data analysis and optimization method under a cloud computing platform according to claim 1, characterized in that: The method for obtaining the garbage to-be-matched area comprises: The garbage matching priority value of all garbage determination areas in each group of garbage determination areas is selected to be the largest, and the corresponding garbage determination area is used as the garbage to-be-matched area.

10. A garbage disposal data analysis and optimization system under a cloud computing platform, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the garbage disposal data analysis and optimization method under the cloud computing platform as described in any one of claims 1 to 9 are implemented.

Images