Urban solid waste monitoring method based on drone and image recognition technology

Abstract

The present invention relates to the field of image analysis technology, and in particular to a method for monitoring urban solid waste based on drones and image recognition technology, the method comprising: obtaining a remote sensing image of a target city and a target image of a garbage area in the target city; determining the similarity between the spectral vector of each area and the spectral vector of the garbage area; determining the texture similarity between each area and the garbage area according to the grayscale co-occurrence matrix of different step lengths of each area and the grayscale co-occurrence matrix constructed at different step lengths of the garbage area; determining the possibility that the area is a garbage area by using the temperature difference and light intensity difference between the remote sensing image and the preset contrast image corresponding to the garbage area, the grayscale co-occurrence matrix constructed at different step lengths of the garbage area, the similarity, and the texture similarity; and determining the area as a garbage area when the possibility is greater than a first threshold. In this way, the present invention improves the accuracy of image analysis of the garbage area where the urban solid waste is located.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and in particular to a method for monitoring urban solid waste based on unmanned aerial vehicles and image recognition technology. Background Art

[0002] With the rapid development of urbanization and population growth, the amount of solid waste generated in cities has continued to rise. Solid waste refers to solid substances that are no longer needed or have no use value and are generated in the course of human activities. These solid wastes may come from a variety of sources such as households, industry, commerce, construction and agriculture. With the development of science and technology, the application of big data and information management in the supervision of solid waste has gradually increased, improving the monitoring and treatment efficiency of solid waste.

[0003] In some scenarios, remote sensing images are often used for image analysis to identify solid waste in cities. However, environmental factors may affect the appearance of solid waste areas in remote sensing images. For example, the spectral reflectance characteristics of objects will change as the ambient temperature changes. In addition, the background environment around solid waste may also cause spectral confusion between solid waste and the background environment, thereby reducing the accuracy of the classification algorithm for image analysis of solid waste. It is easy to mistake other areas for garbage areas where solid waste is located, but solid waste cannot be identified. Therefore, the accuracy of image analysis of garbage areas where solid waste is located in cities using existing methods is low. Summary of the invention

[0004] In order to solve the technical problem of low accuracy of image analysis of garbage areas where urban solid waste is located, the purpose of the present invention is to provide an urban solid waste monitoring method based on drones and image recognition technology. The technical solution adopted is as follows:

[0005] In a first aspect, an embodiment of the present invention provides a method for monitoring urban solid waste based on unmanned aerial vehicles and image recognition technology, comprising: obtaining a remote sensing image of a target city and a target image of a garbage area in the target city through a unmanned aerial vehicle, the remote sensing image and the target image both including images of the target city at different ambient temperatures and different light intensities, and the garbage area including solid waste; determining the similarity between the spectral vector of each area and the spectral vector of the garbage area according to the first spectral characteristics of each area in the remote sensing image and the second spectral characteristics of the garbage area in the target image; determining the texture similarity between each area and the garbage area according to the grayscale co-occurrence matrix of each area with different step sizes and the grayscale co-occurrence matrix constructed at different step sizes of the garbage area; determining the possibility that a region is a garbage area by using the temperature difference and light intensity difference between the remote sensing image and the preset contrast image corresponding to the garbage area, the grayscale co-occurrence matrix constructed at different step sizes of the garbage area, the similarity, and the texture similarity; when the possibility is greater than a first threshold, determining that the region is a garbage area.

[0006] Optionally, when the probability is greater than a first threshold, after determining that the area is a garbage area, the method also includes: determining a first total area of ​​each area in the remote sensing image that is not a garbage area but is identified as a garbage area, determining a second total area of ​​the garbage areas identified in the remote sensing image, determining a third total area of ​​each area in the remote sensing image that is correctly identified as a garbage area, and a fourth total area of ​​the actual garbage areas in the remote sensing image; determining the recognition accuracy based on the first total area, the second total area, the third total area and the fourth total area; when the recognition accuracy is greater than or equal to the second threshold, determining that the recognition accuracy meets the requirement; when the recognition accuracy is less than the second threshold, improving the image quality of the remote sensing image and the target image.

[0007] Optionally, determining the recognition accuracy based on the first total area, the second total area, the third total area and the fourth total area includes: calculating a first ratio of the first total area to the second total area, a second ratio of the third total area to the fourth total area, and a first product between the first ratio and the second ratio; calculating a first difference between a predetermined value and the first product; and normalizing the first difference to obtain the recognition accuracy.

[0008] Optionally, determining the similarity between the spectral vector of each area and the spectral vector of the garbage area based on the first spectral feature of each area of ​​the remote sensing image and the second spectral feature of the garbage area in the target image includes: dividing the remote sensing image into multiple areas and determining the spectral vector of each area; obtaining spectral curves of the garbage area and other areas in the target image, the spectral curves including multiple spectral bands; determining the weights of different spectral bands in identifying the garbage area at each ambient temperature based on the first reflectivity of the garbage area of ​​the target image on the spectral band, the second reflectivity of the adjacent area adjacent to the garbage area on the spectral band, the first pixel value of the garbage area on the spectral band, and the second pixel value of the adjacent area on the spectral band; determining the similarity between the spectral vector of each area and the spectral vector of the garbage area based on the weight, the third reflectivity of each area in the remote sensing image on the spectral band and the vector value of the spectral vector corresponding to the garbage area, the first spectral feature includes the spectral vector of the area, and the second spectral feature includes the spectral vector, the first reflectivity and the spectral band of the garbage area.

[0009] Optionally, determining the weights of different spectral bands at each ambient temperature in identifying the garbage area according to a first reflectivity of the garbage area of ​​the target image on the spectral band, a second reflectivity of an adjacent area adjacent to the garbage area on the spectral band, a first pixel value of the garbage area on the spectral band, and a second pixel value of the adjacent area on the spectral band includes: calculating an absolute value of a third difference between the first reflectivity and the second reflectivity, a first mean value of the first pixel value of the garbage area on the spectral band and a second mean value of the second pixel value of the adjacent area on the spectral band;

[0010] Calculate the absolute value of the fourth difference between the first variance of the first pixel value of the garbage area on the spectral band and the second variance of the second pixel value of the adjacent area on the spectral band; determine the second product of the second difference, the absolute value of the third difference and the absolute value of the fourth difference; superimpose the second products of each area to obtain the weight of the spectral band at the ambient temperature in identifying the garbage area.

[0011] Optionally, determining the similarity between the spectral vector of each region and the spectral vector of the garbage area based on the weight, the third reflectivity of each region in the remote sensing image in the spectral band and the vector value of the spectral vector corresponding to the garbage area includes: calculating the absolute value of the fifth difference between the third reflectivity and the vector value of the spectral vector; determining the third product between the weight and the absolute value of the fifth difference, and superimposing the third products corresponding to each spectral vector of the region to obtain the similarity between the spectral vector of each region and the spectral vector of the garbage area.

[0012] Optionally, determining the texture similarity between each region and the garbage region based on the grayscale co-occurrence matrix of different step sizes of each region and the grayscale co-occurrence matrix constructed at different step sizes of the garbage region includes: constructing the grayscale co-occurrence matrix of different step sizes of each region; determining the first texture richness of the grayscale co-occurrence matrix corresponding to the different step sizes of each region based on the number of types of point pairs in the grayscale co-occurrence matrix of different step sizes of each region and the element value of each element in the grayscale co-occurrence matrix; determining the texture similarity between each region and the garbage region based on the first texture richness, the first energy value and the second energy value, the first energy value being the energy value corresponding to the grayscale co-occurrence matrix constructed at different step sizes of each region, and the second energy value being the energy value corresponding to the grayscale co-occurrence matrix constructed at different step sizes of the garbage region.

[0013] Optionally, determining the texture similarity between each region and the garbage region based on the first texture richness, the first energy value and the second energy value includes: calculating the absolute value of the sixth difference between the second energy value and the first energy value, and the reciprocal of the sum of the absolute value of the sixth difference and a predetermined value; determining the fourth product of the first texture richness and the reciprocal; and determining the average of the fourth products of each step size as the texture similarity between the region and the garbage region.

[0014] Optionally, the possibility of determining a region as a garbage region by using the temperature difference and light intensity difference between the remote sensing image and the preset contrast image corresponding to the garbage region, the grayscale co-occurrence matrix constructed at different step sizes for the garbage region, the similarity and the texture similarity includes: determining the temperature difference and light intensity difference between the remote sensing image and the preset contrast image corresponding to the garbage region; determining the fifth product between the reciprocal of the temperature difference and the reciprocal of the light intensity difference, normalizing the fifth product to obtain a first credibility of the similarity; determining a first texture richness of the grayscale co-occurrence matrix corresponding to different step sizes of each region according to the number of types of point pairs in the grayscale co-occurrence matrix of different step sizes of each region, the grayscale value of each element in the grayscale co-occurrence matrix and the probability of simultaneous occurrence of the grayscale values ​​of two elements; determining a second credibility of the texture similarity according to the first texture richness of each region and the second texture richness of the grayscale co-occurrence matrix constructed at different step sizes for the garbage region; and determining the possibility of a region being a garbage region by using the first credibility, the second credibility, the similarity and the texture similarity.

[0015] Optionally, determining the possibility that an area is a garbage area using the first credibility, the second credibility, the similarity and the texture similarity includes: calculating a first sum between the first credibility and the second credibility, a third ratio between the first credibility and the first sum, and a fourth ratio between the second credibility and the first sum; calculating a sixth product between the third ratio and the similarity and a seventh product between the fourth ratio and the texture similarity; calculating a second sum between the sixth product and the seventh product; and normalizing the second sum to obtain the possibility.

[0016] The invention has the following beneficial effects: firstly, a remote sensing image of a target city and a target image of a garbage area in the target city are obtained by an unmanned aerial vehicle, wherein both the remote sensing image and the target image include images of the target city under different ambient temperatures and different light intensities; then, according to the first spectral features of each area of ​​the remote sensing image and the second spectral features of the garbage area in the target image, the similarity between the spectral vector of each area and the spectral vector of the garbage area is determined; then, according to the gray level co-occurrence matrix of each area with different step sizes and the gray level co-occurrence matrix constructed at different step sizes of the garbage area, the texture similarity between each area and the garbage area is determined; secondly, the possibility that the area is a garbage area is determined by using the temperature difference and the light intensity difference between the remote sensing image and the preset contrast image corresponding to the garbage area, the gray level co-occurrence matrix constructed at different step sizes of the garbage area, the similarity and the texture similarity; finally, when the possibility is greater than a first threshold value, the area is determined to be a garbage area.

[0017] In this way, the embodiment of the present invention can analyze the remote sensing images of the target city under different ambient temperatures and light intensities. Then, based on the similarity between the spectral characteristics of the determined garbage area and the spectral characteristics of each area in the remote sensing image, as well as the texture similarity between each area and the garbage area, determine the possibility that each area in the remote sensing image belongs to the garbage area. Determine whether a certain area in the remote sensing image is a garbage area by comparing the relative size relationship between the possibility and the first threshold. Therefore, the embodiment of the present invention can combine spectral characteristics, texture characteristics and environmental factors to determine the possibility that the area represents a garbage area, and identify the garbage area where the solid waste in the remote sensing image is located based on the possibility. Improve the accuracy of image analysis of garbage areas where solid waste in the city is located. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] 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.

[0019] Figure 1 A flowchart of a method for monitoring urban solid waste based on drones and image recognition technology is provided for one embodiment of the present invention.

[0020] Figure 2 A schematic structural diagram of a municipal solid waste monitoring device based on drone and image recognition technology is provided for one embodiment of the present invention.

[0021] Figure 3A schematic diagram of the structure of a municipal solid waste monitoring system based on drone and image recognition technology provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0022] 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 specific implementation method, structure, characteristics and effects of a method for monitoring urban solid waste based on drone and image recognition technology proposed by the present invention 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.

[0023] 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.

[0024] The following is a detailed description of a method for monitoring urban solid waste based on drones and image recognition technology provided by the present invention in conjunction with the accompanying drawings.

[0025] Embodiment 1:

[0026] See also Figure 1 , which shows a flow chart of a method for monitoring urban solid waste based on drones and image recognition technology provided by an embodiment of the present invention, including:

[0027] S101, obtaining a remote sensing image of a target city and a target image of a garbage area in the target city through a drone.

[0028] Among them, both the remote sensing image and the target image include images of the target city under different ambient temperatures and different light intensities, and the garbage area includes solid waste.

[0029] Specifically, you can choose a suitable drone and a remote sensing sensor, and equip the drone with an infrared camera. Among them, the remote sensing sensor can obtain remote sensing images of the target city. The infrared camera can obtain infrared images of the target city at different ambient temperatures. Further, the software is used to plan the flight route of the drone, set the flight altitude, flight speed and shooting interval of the drone, etc., to ensure that the image overlap rate meets the requirements. And before the flight, check the status of the drone, remote sensing sensor and infrared camera, perform flight missions and monitor the flight status of the drone, and collect real-time image data. Then, the collected images are preliminarily processed, including denoising, geometric correction and radiation correction, to correct the distortion caused by factors such as drone attitude and terrain undulations. Finally, the software is used to stitch multiple images into a complete high-resolution image and perform spectral fusion.

[0030] Furthermore, for the target image of the garbage area in the city, multiple garbage areas can be found in the target city. Under different ambient temperatures and different light intensities, the target image of the garbage area is obtained using a remote sensing sensor carried by a human machine. The target image is a remote sensing image. For the remote sensing image of the target city, the remote sensing image of the target city is obtained using a remote sensing sensor carried by a human machine under different ambient temperatures and different light intensities. The remote sensing image of the target city includes garbage areas and non-garbage areas. More specifically, the spectral curves of the target images of the garbage areas under different ambient temperatures can be obtained at intervals of 2°C, and the same garbage areas under different light intensities at the same temperature are recorded as a group of images. Then the spectral curves are normalized to ensure that the data are on the same scale so that the spectral data collected under different conditions are comparable. The mean of the spectral reflectance of a group of images is calculated as the spectral reflectance of the garbage area under the ambient temperature corresponding to the group of images. The spectral vectors in the same group of images are used for analysis to obtain the importance of different spectral bands in the spectral curves for garbage area image analysis at multiple ambient temperatures.

[0031] S102, determining the similarity between the spectral vector of each region and the spectral vector of the garbage region according to the first spectral feature of each region of the remote sensing image and the second spectral feature of the garbage region in the target image.

[0032] Specifically, the first spectral feature includes but is not limited to the spectral vector, and the second spectral feature includes the spectral vector, spectral reflectance, and spectral band of the garbage area. By comparing the similarity between the spectral vectors of each area and the spectral vector of the garbage area, the area similar to the garbage area can be preliminarily determined, and according to the similarity, the area can be preliminarily determined to be a garbage area.

[0033] Further, when determining the similarity between the spectral vector of each area and the spectral vector of the garbage area, as an optional embodiment of the present invention, the remote sensing image is divided into multiple areas, and the spectral vector of each area is determined; the spectral curves of the garbage area and other areas in the target image are obtained, and the spectral curves include multiple spectral bands; according to the first reflectivity of the garbage area of ​​the target image on the spectral band, the second reflectivity of the adjacent area adjacent to the garbage area on the spectral band, the first pixel value of the garbage area on the spectral band, and the second pixel value of the adjacent area on the spectral band, the weight of different spectral bands in identifying the garbage area at each ambient temperature is determined; according to the weight, the third reflectivity of each area in the remote sensing image on the spectral band and the vector value of the spectral vector corresponding to the garbage area, the similarity between the spectral vector of each area and the spectral vector of the garbage area is determined, the first spectral feature includes the spectral vector of the area, and the second spectral feature includes the spectral vector of the garbage area, the first reflectivity and the spectral band.

[0034] Specifically, for different spectral bands, the greater the difference in spectral reflectance between the garbage area on a certain spectral band and other surrounding areas, the higher the weight that the spectral band should correspond to when distinguishing garbage areas. At the same time, the greater the difference in pixel mean and variance between the garbage area on a certain spectral band and other surrounding areas, the higher the weight that the spectral band should correspond to when distinguishing garbage areas.

[0035] Furthermore, when determining the weights of different spectral bands at various ambient temperatures in identifying garbage areas, as an optional embodiment of the present invention, firstly, the absolute value of the third difference between the second difference between the first reflectivity and the second reflectivity, and the first mean value of the first pixel value of the garbage area on the spectral band and the second mean value of the second pixel value of the adjacent area on the spectral band is calculated;

[0036] Then, the absolute value of the fourth difference between the first variance of the first pixel value of the garbage area on the spectral band and the second variance of the second pixel value of the adjacent area on the spectral band is calculated; then the second product of the second difference, the absolute value of the third difference and the absolute value of the fourth difference is determined; finally, the second products of each area are superimposed to obtain the weight of the spectral band at ambient temperature in identifying the garbage area.

[0037] Specifically, the embodiment of the present invention uses the following formula to calculate the weight of the spectral band at ambient temperature when identifying the garbage area:

[0038]

[0039] In the above formula, express In the target image of the garbage area taken at an ambient temperature of ℃, The weight corresponding to each spectral band. Indicated in In the target image of the garbage area taken at an ambient temperature of The first reflectivity corresponding to the spectral band. Indicated in In the target image of the garbage area taken at an ambient temperature of The adjacent area is The second reflectivity corresponding to the spectral band. Indicated in In the target image of the garbage area taken at an ambient temperature of The first mean of the first pixel values ​​over the spectral bands. Indicated in In the target image of the garbage area taken at an ambient temperature of The adjacent area is A second mean of the second pixel values ​​over the spectral bands. Indicated in In the target image of the garbage area taken at ambient temperature, the garbage area is The first variance of the first pixel value over the spectral band. Indicated in In the target image of the garbage area taken at an ambient temperature of The adjacent area is A second variance of the second pixel value over the spectral band. Indicates adjacent areas. Indicates the number of adjacent regions.

[0040] Among them, As a parameter, it indicates the relationship between the garbage area and the surrounding The difference in pixel values ​​between adjacent areas is used to correct the weights of spectral bands obtained based on different reflectances.

[0041] Furthermore, under the same ambient temperature, the weights of the spectral vector dimensions corresponding to different spectral bands are recorded as:

[0042]

[0043] Furthermore, when the spectral vector of a certain area in the remote sensing image has a high similarity with a garbage area determined in advance at the same ambient temperature under the corresponding ambient temperature, the possibility that the area is a garbage area is also relatively high. When determining the similarity between the spectral vector of each area and the spectral vector of the garbage area, as an optional embodiment of the present invention, firstly calculate the absolute value of the fifth difference between the third reflectivity and the vector value of the spectral vector; then determine the third product between the weight and the absolute value of the fifth difference, and superimpose the third products corresponding to each spectral vector of the area to obtain the similarity between the spectral vector of each area and the spectral vector of the garbage area.

[0044] Specifically, the embodiment of the present invention uses the following formula to calculate the similarity between the spectrum vector of each region and the spectrum vector of the garbage region:

[0045]

[0046] In the above formula, The region segmentation result of the remote sensing image representing the target city The similarity between the spectral vector of the region and the spectral vector of the garbage region at the corresponding ambient temperature when the target image was taken. Indicates the current remote sensing image of the target city corresponding to the ambient temperature of the shooting. The weight corresponding to the spectral vector. Indicated in spectral vector, the regional segmentation result of the remote sensing image of the current target city The third reflectance of the spectral curve corresponding to each region; Indicated in spectral vector dimensions, the vector value of the spectral vector of the garbage area under the ambient temperature value corresponding to the current target image shooting. is the number of spectral vectors.

[0047] S103, determining the texture similarity between each region and the garbage region according to the gray level co-occurrence matrix of each region with different step sizes and the gray level co-occurrence matrix of the garbage region constructed with different step sizes.

[0048] Specifically, the texture of each region can be analyzed based on the gray-level co-occurrence matrix. The gray-level co-occurrence matrix constructed with a smaller step size reflects more detailed texture information, while the gray-level co-occurrence matrix constructed with a larger step size reflects larger-scale texture features. Therefore, it is necessary to analyze the texture features of the region in gray-level co-occurrence matrices with different step sizes. In this way, the texture similarity between each region of the remote sensing image and the known garbage region can be determined.

[0049] Furthermore, when determining the texture similarity between each region and the garbage region, as an optional embodiment of the present invention, a grayscale co-occurrence matrix of different step lengths of each region is first constructed; then, based on the number of types of point pairs in the grayscale co-occurrence matrix of different step lengths of each region and the element value of each element in the grayscale co-occurrence matrix, the first texture richness of the grayscale co-occurrence matrix corresponding to the different step lengths of each region is determined; finally, based on the first texture richness, the first energy value and the second energy value, the texture similarity between each region and the garbage region is determined, the first energy value being the energy value corresponding to the grayscale co-occurrence matrix constructed of each region at different step lengths, and the second energy value being the energy value corresponding to the grayscale co-occurrence matrix constructed of the garbage region at different step lengths.

[0050] Specifically, in the embodiment of the present invention, grayscale co-occurrence matrices with step sizes of 1, 2, 3, 4, and 5 are constructed respectively, and the texture similarity shown by the constructed grayscale co-occurrence matrices is measured at different step sizes. When the grayscale co-occurrence matrix under a certain step size contains richer texture information, it should play a greater reference in the comparison of texture similarity. A higher weight is given to the similarity of the grayscale co-occurrence matrix under the step size. When the number of point pairs in a grayscale co-occurrence matrix is ​​greater, it can be considered that the texture information contained in the grayscale co-occurrence matrix is ​​richer. At the same time, the contrast is used to measure the distribution of the values ​​of the grayscale co-occurrence matrix and the amount of local changes in the remote sensing image, which reflects the clarity of the remote sensing image and the depth of the texture grooves. The deeper the texture grooves, the greater the contrast and the clearer the effect; conversely, the smaller the contrast value, the shallower the grooves and the blurred the effect. In addition, the energy in the grayscale co-occurrence matrix reflects the uniformity of the grayscale distribution and the coarseness of the texture of the remote sensing image. If the element values ​​of the grayscale co-occurrence matrix are closer, the smaller the energy value, indicating a fine texture. If some of the elements have large values ​​and others have small values, the energy value is large, and a large energy value indicates a more uniform and regularly changing texture pattern. The texture similarity between each area and the garbage area is determined by the texture richness and energy value of the point pairs of the gray-level co-occurrence matrix constructed by the length.

[0051] Furthermore, when determining the first texture richness of the gray level co-occurrence matrix corresponding to different step lengths of each region, the present invention specifically uses the following formula for calculation:

[0052]

[0053] In the above formula, Indicates the region segmentation result of the remote sensing image. The region step size is First texture richness of the constructed gray-level co-occurrence matrix. Indicates the step length is The regional segmentation results of remote sensing images The number of point pairs in the gray level co-occurrence matrix of a region. is the normalized gray-level co-occurrence matrix Line The element value of the column element represents the grayscale value and grayscale value The probability of simultaneous occurrence. And in the formula As a whole, the step size is The contrast corresponding to the gray-level co-occurrence matrix of the garbage area. Indicates the number of rows, Indicates the number of columns.

[0054] Furthermore, when determining the texture similarity between each region and the garbage region, as an optional embodiment of the present invention, firstly, the absolute value of the sixth difference between the second energy value and the first energy value, and the reciprocal of the sum of the absolute value of the sixth difference and a predetermined value are calculated; secondly, the fourth product of the first texture richness and the reciprocal is determined; finally, the average of the fourth products of each step size is determined as the texture similarity between the region and the garbage region.

[0055] Specifically, the embodiment of the present invention specifically uses the following formula to calculate the texture similarity between the region and the garbage region:

[0056]

[0057] In the above formula, Indicates the current region segmentation result of the remote sensing image. The texture similarity between the garbage area and the garbage area. Indicates the current region segmentation result of the remote sensing image. The region step size is First texture richness in the constructed gray-level co-occurrence matrix. Represents the known garbage area, with a step size of The second energy value corresponding to the constructed gray-level co-occurrence matrix. Indicates the current region segmentation result of the remote sensing image. regions, the step length is The first energy value corresponding to the constructed gray-level co-occurrence matrix.

[0058] S104, determining the possibility that the area is a garbage area by using the temperature difference and light intensity difference between the remote sensing image and the preset comparison image corresponding to the garbage area, the gray level co-occurrence matrix constructed at different step sizes of the garbage area, the similarity and the texture similarity.

[0059] Specifically, the texture and spectrum of different garbage areas have different representations of the garbage areas, so the similarity and texture similarity in the above embodiments of the present invention have different credibility for identifying garbage areas. For garbage areas, when the temperature difference between the captured remote sensing image and the target image of the known garbage area is smaller, and the light intensity difference with the target image of the known garbage area is smaller, the credibility of identifying the garbage area based on the similarity obtained by the spectral information is higher.

[0060] Further, when determining the possibility that an area is a garbage area, as an optional embodiment of the present invention, first determine the temperature difference and light intensity difference between the remote sensing image and the preset contrast image corresponding to the garbage area; then determine the fifth product between the reciprocal of the temperature difference and the reciprocal of the light intensity difference, normalize the fifth product, and obtain a first credibility of the similarity; then determine the first texture richness of the gray level co-occurrence matrix corresponding to the different step lengths of each area according to the number of types of point pairs in the gray level co-occurrence matrix of different step lengths of each area, the gray value of each element in the gray level co-occurrence matrix, and the probability of simultaneous occurrence of the gray value of each element; secondly, determine the second credibility of the texture similarity according to the first texture richness of each area and the second texture richness of the gray level co-occurrence matrix constructed for the garbage area at different step sizes; finally, use the first credibility, the second credibility, the similarity, and the texture similarity to determine the possibility that the area is a garbage area.

[0061] Specifically, the embodiment of the present invention specifically uses the following formula to calculate the first credibility:

[0062]

[0063] In the above formula, Indicates the measurement of the current remote sensing image. The first confidence level of the similarity between the spectral vectors of a region and a known garbage region. Indicates the temperature difference between the current remote sensing image and the preset comparison image. Indicates the light intensity difference between the current remote sensing image and the preset comparison image. Represents the normalization function, which is used to Perform normalization.

[0064] It is worth noting that in order to avoid and If the formula cannot be calculated due to the value of 0, you can also add a constant 1 to the denominator to avoid this problem. That is, the first credibility can also be calculated using the formula express.

[0065] Furthermore, when the texture expression of the known garbage area and the current area to be evaluated is richer, the information content of the texture expression is higher, and the credibility of the texture similarity is higher. At the same time, when the texture expression of the known garbage area and the area to be evaluated is closer in richness, the texture similarity is more credible. The embodiment of the present invention specifically uses the following formula to calculate the second credibility of texture similarity:

[0066]

[0067] In the above formula, Indicates the measurement of the current remote sensing image. The second confidence level of texture similarity when the area is compared with the known garbage area. Indicates the current region segmentation result. The region step size is First texture richness in the constructed gray-level co-occurrence matrix. Indicates that the step length of the known garbage area is Second texture richness in the constructed gray-level co-occurrence matrix.

[0068] Furthermore, when determining the possibility that a certain area in a remote sensing image is a garbage area, as an optional embodiment of the present invention, firstly, a first sum between the first credibility and the second credibility, a third ratio between the first credibility and the first sum, and a fourth ratio between the second credibility and the first sum are calculated; then, a sixth product between the third ratio and the similarity and a seventh product between the fourth ratio and the texture similarity are calculated; secondly, a second sum between the sixth product and the seventh product is calculated; finally, the second sum is normalized to obtain the possibility.

[0069] Specifically, the embodiment of the present invention specifically uses the following formula to calculate the probability:

[0070]

[0071] In the above formula, Indicates the number of The regions represent the possibility of garbage areas. Indicates the measurement of the current remote sensing image. The first confidence level of the similarity between the spectral vectors of a region and a known garbage region. The region segmentation result of the remote sensing image representing the target city The similarity between the spectral vector of the region and the spectral vector of the garbage region at the corresponding ambient temperature when the target image was taken. Indicates the measurement of the current remote sensing image. The second confidence level of texture similarity when the area is compared with the known garbage area. Indicates the current region segmentation result of the remote sensing image. The texture similarity between the garbage area and the garbage area. Represents the normalization function, which is used to A normalization process is performed, which may specifically be a maximum and minimum value normalization function, and there is no limitation on this.

[0072] S105: When the possibility is greater than the first threshold, determine that the area is a garbage area.

[0073] Specifically, the first threshold can be set according to actual conditions. In the embodiment of the present invention, the first threshold is set to 0.8. When it is greater than 0.8, the first The area is garbage area.

[0074] The embodiment of the present invention can analyze the remote sensing images of the target city under different ambient temperatures and light intensities. Then, based on the similarity between the spectral characteristics of the determined garbage area and the spectral characteristics of each area in the remote sensing image, as well as the texture similarity between each area and the garbage area, determine the possibility that each area in the remote sensing image belongs to the garbage area. By comparing the relative size relationship between the possibility and the first threshold, determine whether a certain area in the remote sensing image is a garbage area. Therefore, the embodiment of the present invention can combine spectral characteristics, texture characteristics and environmental factors to determine the possibility that the area represents a garbage area, and identify the garbage area where the solid waste in the remote sensing image is located based on the possibility. Improve the accuracy of image analysis of garbage areas where solid waste in the city is located.

[0075] Furthermore, the embodiment of the present invention can also evaluate the recognition accuracy of the garbage area identified by the above method. First, use remote sensing image processing software to view the high-resolution remote sensing images taken by the drone. Manually analyze the features in the remote sensing image to find typical manifestations of domestic garbage, such as irregular shapes, unnatural colors, and obvious contrast with the surrounding environment. Divide the suspicious areas based on the observed features. Manually outline these areas and create location marks for domestic garbage. In order to improve the accuracy of recognition, these marks are checked multiple times and compared with the results of on-site field investigations to ensure the correctness of recognition. When the method of the embodiment of the present invention is used to identify garbage areas, if more areas are mistakenly identified as garbage areas and more garbage areas are not identified, it means that the accuracy of the method of the embodiment of the present invention is low. Then compare the actual garbage areas manually screened above with the garbage areas identified by the method of the embodiment of the present invention to evaluate the recognition accuracy provided by the above embodiment of the present invention.

[0076] Further, when evaluating the recognition accuracy, as an optional embodiment of the present invention, a first total area of ​​each area in the remote sensing image that is not a garbage area but is identified as a garbage area is determined, a second total area of ​​the garbage areas identified in the remote sensing image is determined, a third total area of ​​each area in the remote sensing image that is correctly identified as a garbage area, and a fourth total area of ​​the actual garbage areas in the remote sensing image are determined; the recognition accuracy is determined based on the first total area, the second total area, the third total area, and the fourth total area; when the recognition accuracy is greater than or equal to a second threshold, it is determined that the recognition accuracy meets the requirement; when the recognition accuracy is less than the second threshold, the image quality of the remote sensing image and the target image is improved.

[0077] Specifically, the second threshold in the embodiment of the present invention can be determined according to actual conditions, and the value is 0.8 in the embodiment of the present invention. Furthermore, when determining the recognition accuracy, firstly, a first ratio of the first total area to the second total area, a second ratio of the third total area to the fourth total area, and a first product between the first ratio and the second ratio are calculated; then a first difference between the predetermined value and the first product is calculated; finally, the first difference is normalized to obtain the recognition accuracy.

[0078] Specifically, the embodiment of the present invention specifically uses the following formula to calculate the recognition accuracy:

[0079]

[0080] In the above formula, It represents the recognition accuracy of the above process of the embodiment of the present invention. It represents the first total area which is not a garbage area in the remote sensing image but is identified as a garbage area in the above process of the embodiment of the present invention. It represents the second total area of ​​the garbage area identified by the above process in the embodiment of the present invention. It represents the third total area of ​​the correctly identified garbage area in the above process of the embodiment of the present invention. Represents the fourth total area of ​​garbage areas in real life. Represents the normalization function, which is used to Perform normalization.

[0081] When the recognition accuracy It is believed that the accuracy of identifying garbage areas using the method provided by the embodiment of the present invention meets the requirements. When , it is considered that the requirements are not met. At this time, the quality of the acquired image can be improved and more reference images can be obtained.

[0082] Embodiment 2:

[0083] Based on the urban solid waste monitoring method based on drone and image recognition technology provided in the above embodiment, based on the same technical concept, an embodiment of the present invention also provides an urban solid waste monitoring device based on drone and image recognition technology. Figure 2 A schematic diagram of the structure of a municipal solid waste monitoring device based on drone and image recognition technology is provided in one embodiment of the present invention. Figure 2 As shown. The urban solid waste monitoring device 200 based on drone and image recognition technology includes: an acquisition module 201, which is used to acquire a remote sensing image of a target city and a target image of a garbage area in the target city through a drone, wherein the remote sensing image and the target image both include images of the target city at different ambient temperatures and different light intensities, and the garbage area includes solid waste; a determination module 202, which is used to determine the similarity between the spectral vector of each area and the spectral vector of the garbage area according to the first spectral feature of each area of ​​the remote sensing image and the second spectral feature of the garbage area in the target image; the determination module 202 is also used to determine the texture similarity between each area and the garbage area according to the gray level co-occurrence matrix of different step lengths of each area and the gray level co-occurrence matrix constructed at different step lengths of the garbage area; the determination module 202 is also used to determine the possibility that the area is a garbage area by using the temperature difference and light intensity difference between the remote sensing image and the preset contrast image corresponding to the garbage area, the gray level co-occurrence matrix constructed at different step lengths of the garbage area, the similarity and the texture similarity; the determination module 202 is also used to determine that the area is a garbage area when the possibility is greater than the first threshold.

[0084] The embodiment of the present invention can analyze the remote sensing images of the target city under different ambient temperatures and light intensities. Then, based on the similarity between the spectral characteristics of the determined garbage area and the spectral characteristics of each area in the remote sensing image, as well as the texture similarity between each area and the garbage area, determine the possibility that each area in the remote sensing image belongs to the garbage area. By comparing the relative size relationship between the possibility and the first threshold, determine whether a certain area in the remote sensing image is a garbage area. Therefore, the embodiment of the present invention can combine spectral characteristics, texture characteristics and environmental factors to determine the possibility that the area represents a garbage area, and identify the garbage area where the solid waste in the remote sensing image is located based on the possibility. Improve the accuracy of image analysis of garbage areas where solid waste in the city is located.

[0085] Optionally, the determination module 202 is further used to determine a first total area of ​​each area in the remote sensing image that is not a garbage area but is identified as a garbage area, determine a second total area of ​​the garbage areas identified in the remote sensing image, determine a third total area of ​​each area in the remote sensing image that is correctly identified as a garbage area, and a fourth total area of ​​the actual garbage areas in the remote sensing image; determine the recognition accuracy based on the first total area, the second total area, the third total area and the fourth total area; when the recognition accuracy is greater than or equal to a second threshold, determine that the recognition accuracy meets the requirement; when the recognition accuracy is less than the second threshold, improve the image quality of the remote sensing image and the target image.

[0086] Optionally, the determination module 202 is also used to calculate a first ratio of the first total area to the second total area, a second ratio of the third total area to the fourth total area, and a first product between the first ratio and the second ratio; calculate a first difference between a predetermined value and the first product; and normalize the first difference to obtain recognition accuracy.

[0087] Optionally, the determination module 202 is further used to divide the remote sensing image into multiple regions and determine the spectral vector of each region; obtain spectral curves of the garbage region and other regions in the target image, the spectral curves including multiple spectral bands; determine the weights of different spectral bands in identifying the garbage region at each ambient temperature according to the first reflectivity of the garbage region of the target image on the spectral band, the second reflectivity of the adjacent region adjacent to the garbage region on the spectral band, the first pixel value of the garbage region on the spectral band, and the second pixel value of the adjacent region on the spectral band; determine the similarity between the spectral vector of each region and the spectral vector of the garbage region according to the weight, the third reflectivity of each region in the remote sensing image on the spectral band and the vector value of the spectral vector corresponding to the garbage region, the first spectral feature including the spectral vector of the region, and the second spectral feature including the spectral vector of the garbage region, the first reflectivity and the spectral band.

[0088] Optionally, the determination module 202 is further configured to calculate an absolute value of a second difference between the first reflectivity and the second reflectivity, and a third difference between a first mean value of a first pixel value of the garbage area on the spectral band and a second mean value of a second pixel value of an adjacent area on the spectral band;

[0089] Calculate the absolute value of the fourth difference between the first variance of the first pixel value of the garbage area on the spectral band and the second variance of the second pixel value of the adjacent area on the spectral band; determine the second product of the second difference, the absolute value of the third difference and the absolute value of the fourth difference; superimpose the second products of each area to obtain the weight of the spectral band at the ambient temperature in identifying the garbage area.

[0090] Optionally, the determination module 202 is also used to calculate the absolute value of the fifth difference between the third reflectivity and the vector value of the spectral vector; determine the third product between the weight and the absolute value of the fifth difference, and superimpose the third products corresponding to each spectral vector of the region to obtain the similarity between the spectral vector of each region and the spectral vector of the garbage area.

[0091] Optionally, the determination module 202 is also used to construct grayscale co-occurrence matrices of different step sizes for each region; determine the first texture richness of the grayscale co-occurrence matrices corresponding to the different step sizes of each region according to the number of types of point pairs in the grayscale co-occurrence matrices of different step sizes for each region and the element value of each element in the grayscale co-occurrence matrix; determine the texture similarity between each region and the garbage region according to the first texture richness, the first energy value and the second energy value, the first energy value being the energy value corresponding to the grayscale co-occurrence matrix constructed for each region at different step sizes, and the second energy value being the energy value corresponding to the grayscale co-occurrence matrix constructed for the garbage region at different step sizes.

[0092] Optionally, the determination module 202 is also used to calculate the absolute value of the sixth difference between the second energy value and the first energy value, and the reciprocal of the sum of the absolute value of the sixth difference and a predetermined value; determine the fourth product of the first texture richness and the reciprocal; and determine the average of the fourth products of each step size as the texture similarity between the region and the garbage region.

[0093] Optionally, the determination module 202 is also used to determine the temperature difference and light intensity difference between the remote sensing image and the preset contrast image corresponding to the garbage area; determine the fifth product between the inverse of the temperature difference and the inverse of the light intensity difference, normalize the fifth product, and obtain a first credibility of the similarity; determine the first texture richness of the gray level co-occurrence matrix corresponding to the different step lengths of each area according to the number of types of point pairs in the gray level co-occurrence matrix of different step lengths of each area, the gray value of each element in the gray level co-occurrence matrix, and the probability of simultaneous occurrence of the gray value of each element; determine the second credibility of texture similarity according to the first texture richness of each area and the second texture richness of the gray level co-occurrence matrix constructed for the garbage area at different step sizes; use the first credibility, the second credibility, the similarity and the texture similarity to determine the possibility that the area is a garbage area.

[0094] Optionally, the determination module 202 is also used to calculate a first sum between the first credibility and the second credibility, a third ratio between the first credibility and the first sum, and a fourth ratio between the second credibility and the first sum; calculate a sixth product between the third ratio and the similarity and a seventh product between the fourth ratio and the texture similarity; calculate a second sum between the sixth product and the seventh product; and normalize the second sum to obtain the possibility.

[0095] Embodiment three:

[0096] Corresponding to the urban solid waste monitoring method based on drone and image recognition technology provided in the above embodiment, based on the same technical concept, an embodiment of the present invention also provides an urban solid waste monitoring system based on drone and image recognition technology, and the urban solid waste monitoring system based on drone and image recognition technology is used to execute the above urban solid waste monitoring method based on drone and image recognition technology. Figure 3 A schematic diagram of a system for monitoring urban solid waste based on drones and image recognition technology is provided in accordance with an embodiment of the present invention. Figure 3 The urban solid waste monitoring system based on drone and image recognition technology may have relatively large differences due to different configurations or performances, and may include one or more processors 301 and memory 302, the memory 302 is used to store computer programs that can be run on the processor 301, and the processor 301 is used to execute the program stored in the memory 302 to achieve the above Figure 1 The various steps in the method embodiment. The memory 302 may be a temporary storage or a permanent storage. The application stored in the memory 302 may include one or more modules (not shown in the figure), each of which may include a series of computer executable instructions in the urban solid waste monitoring system based on drones and image recognition technology.

[0097] Furthermore, the processor 301 can be configured to communicate with the memory 302 to execute a series of computer executable instructions in the memory 302 on the urban solid waste monitoring system based on drones and image recognition technology. The urban solid waste monitoring system based on drones and image recognition technology can also include one or more power supplies 303, one or more wired or wireless network interfaces 304, one or more input and output interfaces 305, and one or more keyboards 306.

[0098] Specifically in this embodiment, the urban solid waste monitoring system based on drone and image recognition technology includes a processor, a communication interface, a memory and a communication bus; wherein the processor, the communication interface and the memory communicate with each other through the bus; the memory is used to store computer programs; the processor is used to execute the programs stored in the memory to achieve the above Figure 1 The various steps in the method embodiment are similar to those in the method embodiment, and have the beneficial effects of the above method embodiments. To avoid repetition, the embodiments of the present invention will not be described in detail here.

[0099] It should be noted that the urban solid waste monitoring system based on drone and image recognition technology provided in the embodiment of the present invention and the urban solid waste monitoring method based on drone and image recognition technology provided in the embodiment of the present invention are based on the same application concept. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned urban solid waste monitoring method based on drone and image recognition technology, and has the same or similar beneficial effects, and the repetitions will not be repeated.

[0100] It should be noted that the sequence of the above embodiments of the present invention is for description only 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.

[0101] 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 urban solid waste based on drones and image recognition technology, characterized in that: The urban solid waste monitoring method based on drone and image recognition technology includes: Acquire a remote sensing image of a target city and a target image of a garbage area in the target city by using a drone, wherein the remote sensing image and the target image both include images of the target city at different ambient temperatures and different light intensities, and the garbage area includes solid waste; Determining the similarity between the spectral vector of each of the regions and the spectral vector of the garbage region according to the first spectral feature of each region of the remote sensing image and the second spectral feature of the garbage region in the target image; Determining the texture similarity between each of the regions and the garbage region according to the gray level co-occurrence matrices of the regions with different step sizes and the gray level co-occurrence matrices of the garbage region constructed at different step sizes; Determine the possibility that the area is a garbage area by using the temperature difference and light intensity difference between the remote sensing image and a preset contrast image corresponding to the garbage area, the gray level co-occurrence matrix constructed at different step sizes for the garbage area, the similarity and the texture similarity; When the possibility is greater than a first threshold, determining the area as a garbage area; The method further comprises: Determine a first total area of ​​each of the regions in the remote sensing image that is not a garbage region but is identified as a garbage region, determine a second total area of ​​garbage regions identified in the remote sensing image, determine a third total area of ​​each region in the remote sensing image that is correctly identified as a garbage region, and determine a fourth total area of ​​actual garbage regions in the remote sensing image; determining recognition accuracy according to the first total area, the second total area, the third total area, and the fourth total area; When the recognition accuracy is greater than or equal to a second threshold, determining that the recognition accuracy meets the requirement; When the recognition accuracy is less than the second threshold, the image quality of the remote sensing image and the target image is improved.

2. The method for monitoring urban solid waste based on drone and image recognition technology according to claim 1, characterized in that: The determining of the recognition accuracy according to the first total area, the second total area, the third total area and the fourth total area comprises: calculating a first ratio of the first total area to the second total area, a second ratio of the third total area to the fourth total area, and a first product of the first ratio and the second ratio; calculating a first difference between a predetermined value and the first product; The first difference is normalized to obtain the recognition accuracy.

3. The urban solid waste monitoring method based on drone and image recognition technology according to claim 1 is characterized in that: Determining the similarity between the spectral vector of each region and the spectral vector of the garbage region according to the first spectral feature of each region of the remote sensing image and the second spectral feature of the garbage region in the target image comprises: Segmenting the remote sensing image into a plurality of regions and determining a spectral vector of each of the regions; Acquire spectral curves of the garbage area and other areas in the target image, wherein the spectral curves include multiple spectral bands; Determine the weights of different spectral bands at different ambient temperatures in identifying garbage areas according to the first reflectivity of the garbage area of ​​the target image on the spectral band, the second reflectivity of an adjacent area adjacent to the garbage area on the spectral band, the first pixel value of the garbage area on the spectral band, and the second pixel value of the adjacent area on the spectral band; The similarity between the spectral vector of each area and the spectral vector of the garbage area is determined based on the weight, the third reflectivity of each area in the remote sensing image in the spectral band and the vector value of the spectral vector corresponding to the garbage area, the first spectral feature includes the spectral vector of the area, and the second spectral feature includes the spectral vector of the garbage area, the first reflectivity and the spectral band.

4. The urban solid waste monitoring method based on drone and image recognition technology according to claim 3 is characterized in that: The step of determining the weights of different spectral bands at different ambient temperatures when identifying garbage areas according to the first reflectivity of the garbage area of ​​the target image at different ambient temperatures on the spectral band, the second reflectivity of an adjacent area adjacent to the garbage area on the spectral band, the first pixel value of the garbage area on the spectral band, and the second pixel value of the adjacent area on the spectral band comprises: Calculate the absolute value of a second difference between the first reflectivity and the second reflectivity, and a third difference between a first mean value of first pixel values ​​of the garbage area in the spectral band and a second mean value of second pixel values ​​of the adjacent area in the spectral band; Calculate an absolute value of a fourth difference between a first variance of a first pixel value of the garbage area on the spectral band and a second variance of a second pixel value of the adjacent area on the spectral band; determining a second product of the second difference, the absolute value of the third difference, and the absolute value of the fourth difference; The second products of the regions are superimposed to obtain the weight of the spectral band at the ambient temperature in identifying the garbage region.

5. The method for monitoring urban solid waste based on drone and image recognition technology according to claim 3 is characterized in that: Determining the similarity between the spectral vector of each area and the spectral vector of the garbage area according to the weight, the third reflectivity of each area in the remote sensing image in the spectral band, and the vector value of the spectral vector corresponding to the garbage area includes: calculating an absolute value of a fifth difference between the third reflectivity and the vector value of the spectral vector; A third product between the weight and the absolute value of the fifth difference is determined, and the third products corresponding to the spectral vectors of the regions are superimposed to obtain similarities between the spectral vectors of the regions and the spectral vector of the garbage region.

6. The method for monitoring urban solid waste based on drone and image recognition technology according to claim 1, characterized in that: Determining the texture similarity between each of the regions and the garbage region according to the gray level co-occurrence matrix of the regions with different step sizes and the gray level co-occurrence matrix of the garbage region constructed at different step sizes includes: Constructing gray level co-occurrence matrices of different step lengths for each of the regions; Determine the first texture richness of the gray level co-occurrence matrix corresponding to the different step lengths of each of the regions according to the number of types of point pairs in the gray level co-occurrence matrix of the different step lengths of each of the regions and the element value of each element in the gray level co-occurrence matrix; According to the first texture richness, the first energy value and the second energy value, the texture similarity between each of the regions and the garbage region is determined, the first energy value is the energy value corresponding to the gray-level co-occurrence matrix constructed for each of the regions at different step sizes, and the second energy value is the energy value corresponding to the gray-level co-occurrence matrix constructed for the garbage region at different step sizes.

7. The method for monitoring urban solid waste based on drone and image recognition technology according to claim 6 is characterized in that: The determining, according to the first texture richness, the first energy value, and the second energy value, of the texture similarity between each of the regions and the garbage region comprises: calculating an absolute value of a sixth difference between the second energy value and the first energy value, and a reciprocal of a sum of the absolute value of the sixth difference and a predetermined value; determining a fourth product of the first texture richness and the reciprocal; The average value of the fourth product of each step size is determined as the texture similarity between the region and the garbage region.

8. The method for monitoring urban solid waste based on drone and image recognition technology according to claim 1, characterized in that: The possibility of determining that the area is a garbage area by using the temperature difference and light intensity difference between the remote sensing image and the preset contrast image corresponding to the garbage area, the gray level co-occurrence matrix constructed at different step sizes for the garbage area, the similarity and the texture similarity includes: Determining a temperature difference and a light intensity difference between the remote sensing image and a preset comparison image corresponding to the garbage area; determining a fifth product between the reciprocal of the temperature difference and the reciprocal of the light intensity difference, and normalizing the fifth product to obtain a first credibility of the similarity; Determine the first texture richness of the gray level co-occurrence matrix corresponding to the different step lengths of each of the regions according to the number of types of point pairs in the gray level co-occurrence matrix of the different step lengths of each of the regions, the gray value of each element in the gray level co-occurrence matrix, and the probability of simultaneous occurrence of the gray values ​​of two elements; Determining a second credibility of the texture similarity according to the first texture richness of each of the regions and the second texture richness of the gray level co-occurrence matrix constructed in the garbage region at different step sizes; The possibility that the region is a garbage region is determined by using the first credibility, the second credibility, the similarity, and the texture similarity.

9. The method for monitoring urban solid waste based on drone and image recognition technology according to claim 8, characterized in that: The possibility of determining that the region is a garbage region by using the first credibility, the second credibility, the similarity and the texture similarity includes: calculating a first sum value between the first credibility and the second credibility, a third ratio between the first credibility and the first sum value, and a fourth ratio between the second credibility and the first sum value; calculating a sixth product between the third ratio and the similarity and a seventh product between the fourth ratio and the texture similarity; calculating a second sum between the sixth product and the seventh product; The second sum value is normalized to obtain the possibility.

Images