An image recognition-based dusting state monitoring control method

Abstract

The present application relates to the technical field of image recognition, and particularly relates to a lime dusting state monitoring control method based on image recognition. After the lime dusting is completed, the camera is controlled to scan the disinfection surface along the guide rail and shoot a state video, which is uploaded to a cloud server. The server obtains non-overlapping area images by calculating the interval of screenshots, and classifies the images into evaluation graph groups corresponding to the non-overlapping area and the overlapping area of the nozzle according to a preset segmentation path. Then, based on a color feature extraction algorithm, the coverage ratio of lime powder in each evaluation graph group is calculated, and by analyzing the average value, standard deviation and cause-effect relationship with adjacent areas, the lime powder spraying abnormality index corresponding to each nozzle is comprehensively calculated, so as to accurately evaluate the lime dusting quality and locate the abnormal nozzle. The present application realizes the automatic and accurate monitoring and abnormal diagnosis of the lime dusting quality, and effectively improves the reliability and efficiency of the reeling stand disinfection.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and more specifically, to a method for monitoring and controlling the state of ash spreading based on image recognition. Background Technology

[0002] In modern sericulture, disinfection of the silkworm rearing platform is a crucial step in preventing diseases and ensuring the yield and quality of silkworm cocoons. Lime powder, due to its highly efficient disinfection and drying properties, is widely used for disinfecting silkworm rearing platforms. Traditional automated lime-spraying systems typically use preset programs to control the movement of a sprayer above the platform, evenly spraying lime powder onto the green disinfection surface.

[0003] However, existing lime-spreading systems lack effective quality monitoring mechanisms. After the lime-spreading operation is completed, whether its quality meets standards (such as whether the lime coverage is uniform, whether there are any missed areas or blockages) often relies on manual inspection. This method has significant drawbacks: First, manual inspection is highly subjective, inefficient, and difficult to detect subtle unevenness issues; second, when poor lime-spreading results are found, it is impossible to quickly and accurately pinpoint the cause of the abnormality (e.g., which specific nozzle is blocked or the powder output is uneven), leading to time-consuming and labor-intensive troubleshooting and repair, potentially delaying the optimal disinfection time and posing potential risks to aquaculture production. Summary of the Invention

[0004] The purpose of this invention is to provide an image recognition-based method for monitoring and controlling the ash-spreading status, so as to solve the above-mentioned technical problems.

[0005] To achieve the above objectives, the embodiments of this application provide the following technical solutions:

[0006] This application provides an image recognition-based method for monitoring and controlling the state of lime spreading. The method is applicable to a silkworm rearing disinfection system. The system includes a square-shaped disinfection surface, a spraying frame, and a camera. The spraying frame and camera move along the long side of the disinfection surface via a guide rail above it. The spraying frame has multiple nozzles evenly distributed on it, which spray lime powder from a storage hopper in a fan shape onto the disinfection surface. The disinfection surface is green. The method includes: one minute after the lime spreading operation is completed, the camera moves along the guide rail at a preset speed above the disinfection surface, continuously capturing a video of the surface's state. This video is then sent to a cloud-based server to assess the quality of the lime spreading operation. In case of any abnormality, the server determines the cause of the abnormality based on the video of the disinfection surface's state. The lime spreading operation is the spraying of lime powder by the spraying frame along the guide rail onto the disinfection surface.

[0007] Optionally, the PTZ server evaluates the ash-spreading quality of this ash-spreading operation, including:

[0008] The guide rail running speed is obtained, and the corresponding screenshot interval is retrieved from the running parameter reference table based on the guide rail running speed. This results in multiple screenshots of the status video arranged in a time series, and the disinfection area of ​​the silkworm platform corresponding to each screenshot does not overlap.

[0009] Multiple screenshots are segmented according to a preset segmentation path to obtain an evaluation image group corresponding to each segmentation path. The lime powder spraying anomaly index corresponding to the evaluation image group is analyzed, and the lime powder spraying quality of the nozzle is evaluated based on the lime powder spraying anomaly index.

[0010] Optionally, multiple screenshots are segmented according to a preset segmentation path to obtain an evaluation image set corresponding to each segmentation path, including:

[0011] Based on the preset viewfinder dividing line, each screenshot is divided into multiple first screenshot sub-regions and multiple second screenshot sub-regions, wherein the first screenshot sub-regions and the second screenshot sub-regions are interspersed. The first screenshot sub-regions represent the non-overlapping areas of lime spraying, and the second screenshot sub-regions represent the overlapping areas of lime spraying.

[0012] Multiple first screenshot sub-regions are classified according to a preset segmentation path to obtain multiple first screenshot sub-regions corresponding to each nozzle non-overlapping area. These sub-regions are then arranged sequentially according to the nozzle's movement path to obtain a first evaluation image group corresponding to each nozzle non-overlapping area.

[0013] Multiple second screenshot sub-regions are classified according to a preset segmentation path, thereby obtaining multiple second screenshot sub-regions corresponding to the overlapping areas of two adjacent nozzles. These sub-regions are then arranged sequentially according to the movement path of the nozzles, thus obtaining the second evaluation image group corresponding to the overlapping areas of two adjacent nozzles.

[0014] Optionally, analyze and evaluate the lime powder spraying anomaly index corresponding to the image group, including:

[0015] A color-based image feature extraction algorithm sequentially evaluates the first lime spraying anomaly frequency index of the nozzles corresponding to each first evaluation image group.

[0016] The color-based image feature extraction algorithm sequentially evaluates the second lime spraying abnormality frequency index and the third lime spraying abnormality frequency index of two adjacent nozzles corresponding to each second evaluation image group.

[0017] The system calculates the first, second, and third lime spraying abnormality frequency indices for each nozzle, and uses a weighted algorithm to comprehensively evaluate the nozzle's operating condition index. If the operating condition index is less than a preset threshold, the system determines that the nozzle is spraying abnormally and generates a corresponding abnormality warning command so that staff can adjust or replace the abnormal nozzle in a timely manner.

[0018] Optionally, the color-based image feature extraction algorithm sequentially evaluates the first lime spraying anomaly frequency index of the nozzle corresponding to each first evaluation image group, including:

[0019] The first screenshot sub-region in the first evaluation image group is obtained and divided into multiple lime distribution evaluation regions. White pixels in each lime distribution evaluation region are extracted based on a color-based image recognition algorithm. The proportion of white pixels in the total pixels of the lime distribution evaluation region is calculated to obtain the lime powder coverage ratio in each lime distribution evaluation region. The average coverage ratio of lime powder and the standard deviation of multiple coverage ratios are calculated for the multiple lime distribution evaluation regions in the first screenshot sub-region.

[0020] Based on the average coverage ratio and standard deviation, the spraying quality index corresponding to the first screenshot sub-region is calculated by a weighted algorithm. If the spraying quality index exceeds the threshold, the first screenshot sub-region is marked as the first abnormal spraying. The number of first abnormal sprayings in the first evaluation map group is counted, and then the first abnormal frequency of the nozzle corresponding to the first evaluation map is calculated, which is denoted as the first lime spraying abnormal frequency index.

[0021] Optionally, the color-based image feature extraction algorithm sequentially evaluates the second and third lime spraying anomaly frequency indices of two adjacent nozzles corresponding to each second evaluation image group, including:

[0022] The color-based image feature extraction algorithm calculates the second and third lime spraying anomaly frequency indices of the two adjacent nozzles corresponding to each second evaluation image group, including:

[0023] The second screenshot sub-region in the second evaluation image group is obtained and divided into multiple lime distribution evaluation areas. White pixels in each lime distribution evaluation area are extracted based on a color-based image recognition algorithm. The proportion of white pixels in the total pixels of the lime distribution evaluation area is calculated to obtain the lime powder coverage ratio in each lime distribution evaluation area. The average lime powder coverage ratio corresponding to multiple lime distribution evaluation areas in the first screenshot sub-region is also calculated.

[0024] If the average coverage ratio is 50%-70% of the ideal average coverage value, then retrieve the average coverage ratios of the two first screenshot sub-regions adjacent to the second screenshot sub-region, and add a second lime spraying anomaly to the nozzle corresponding to the first screenshot sub-region with the larger average coverage ratio.

[0025] If the average coverage ratio is 5%-49% of the ideal average coverage value, then retrieve the average coverage ratios of the two first screenshot sub-regions adjacent to the second screenshot sub-region, and add a second lime spraying anomaly to the nozzle corresponding to the first screenshot sub-region with the larger average coverage ratio.

[0026] Furthermore, if the average coverage ratio is smaller than the ideal average coverage value, the nozzle corresponding to the first screenshot sub-area with the smaller average coverage ratio will be subject to a third lime spraying anomaly.

[0027] The number of times the second lime spraying abnormality occurred and the number of times the third lime spraying abnormality occurred for each nozzle were counted, and then the frequency index of the second lime spraying abnormality and the frequency index of the third lime spraying abnormality for each nozzle were obtained.

[0028] The beneficial effects of this invention are as follows:

[0029] The image recognition-based ash-spreading status monitoring and control method described in this invention uses a camera to automatically cruise and capture images after ash is spread, and utilizes a PTZ server for image analysis, completely replacing the inefficient and subjective manual inspection mode, and significantly improving the efficiency and objectivity of monitoring.

[0030] Specifically, by using image segmentation and analysis methods, the image of the entire disinfection surface is decomposed into evaluation image groups corresponding to each nozzle and its superimposed area. This method can not only discover the overall phenomenon of uneven dust distribution, but also accurately locate the specific single or multiple nozzles that are malfunctioning. It achieves a leap from "discovering the problem" to "locating the root cause of the problem", providing clear guidance for staff to perform rapid maintenance.

[0031] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a schematic diagram of a method for monitoring and controlling the ash-spreading status based on image recognition, as described in an embodiment of the present invention.

[0034] Figure 2 This is a distribution diagram of the superimposed and non-superimposed areas of the fan-shaped spray patterns corresponding to the multiple nozzles described in this embodiment of the invention;

[0035] Figure 3 The cut-off section described in the embodiments of the present invention Figure 1 The distribution map of the segmentation path to screenshot N. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0037] It should be noted that similar reference numerals or letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0038] Example 1:

[0039] like Figure 1As shown, this embodiment provides an image recognition-based method for monitoring and controlling the state of lime spreading. The method is applicable to a silkworm rearing disinfection system. The system includes a silkworm rearing disinfection surface arranged in a rectangular shape, a spraying frame, and a camera. The spraying frame and camera move along the long side of the silkworm rearing disinfection surface via a guide rail above the surface. The spraying frame has multiple nozzles evenly distributed on it. These nozzles spray lime powder from the storage hopper in a fan shape onto the disinfection surface. The disinfection surface is green, while the lime powder is white; therefore, they can be distinguished using a color-based image feature extraction algorithm. The method includes:

[0040] Step S100: One minute after the ash-spraying operation is completed, the camera passes over the silkworm rearing disinfection surface at a preset speed along the guide rail, continuously recording the status video of the silkworm rearing disinfection surface and sending the status video to the cloud platform server so that the cloud platform server can evaluate the ash-spraying quality of this ash-spraying operation and determine the abnormal factors based on the status video of the silkworm rearing disinfection surface in case of ash-spraying abnormality. The ash-spraying operation is the operation of the spraying frame spraying lime powder onto the silkworm rearing disinfection surface through the guide rail.

[0041] Secondly, in this implementation, the PTZ server evaluates the quality of the ash-spreading operation, including:

[0042] Step S110: Obtain the guide rail running speed, and retrieve the corresponding screenshot interval time from the running parameter reference table based on the guide rail running speed, thereby obtaining multiple screenshots corresponding to the status video arranged in time sequence, and the silkworm platform disinfection area corresponding to each screenshot does not overlap. It should be noted that the above camera is a high-speed camera, so as to make the screenshot as clear as possible and reduce the screenshot ghosting.

[0043] Step S120, as follows Figure 2 - Figure 3 As shown, multiple screenshots are segmented according to a preset segmentation path to obtain an evaluation image group corresponding to each segmentation path. The lime powder spraying anomaly index corresponding to the evaluation image group is analyzed, and the lime powder spraying quality of the nozzle is evaluated based on the lime powder spraying anomaly index.

[0044] The ideal spray pattern of multiple nozzles arranged side-by-side on a spray frame is a fan-shaped distribution, with more nozzles in the center and fewer at the edges. This is determined by the structure of the fan-shaped nozzles themselves. Therefore, when designing the nozzle spacing, the edges of the fan-shaped spray pattern corresponding to each nozzle are designed to intersect to a certain extent, thus forming a pattern like... Figure 2 The superimposed and non-superimposed areas are shown;

[0045] like Figure 3 As shown, this is a cut-off Figure 1 The distribution map of the segmentation paths to screenshot N. Figure 3 The three dots in the middle are an abbreviated meaning used to represent a truncated section. Figure 1 The multiple screenshots between screenshot N are arranged sequentially according to the path, and there may be a certain gap between the corresponding silkworm platform disinfection areas of the multiple screenshots described in this embodiment. Therefore, the multiple sub-regions of the multiple screenshots on the same segmentation path can indirectly reflect the lime powder spraying situation of a nozzle. That is, the analysis and evaluation chart group can indirectly reflect the lime powder spraying situation of a nozzle.

[0046] The analysis and evaluation of this representation of the image group also needs to distinguish between the superimposed area and the non-superimposed area. The non-superimposed area corresponds to the central area of ​​the fan-shaped spray of a single nozzle, while the superimposed area corresponds to the edge area of ​​the fan-shaped spray of two adjacent nozzles.

[0047] Secondly, the early abnormalities of the nozzles are sporadic. Therefore, the core of this embodiment is to detect the abnormal frequency of each nozzle on the corresponding spraying path. When the abnormal frequency is too high, we will intervene in advance to adjust or replace the nozzle, rather than waiting for the nozzle to show obvious abnormalities. At this time, multiple obvious blank areas will appear on the silkworm platform disinfection surface, which has already seriously affected the lime powder spraying operation.

[0048] Since the fan-shaped spray through a nozzle consists of one non-overlapping zone and two overlapping zones, the frequency of anomalies in the non-overlapping zone and the frequency of anomalies in the overlapping zone have different weightings in the anomaly evaluation system. For details on the specific evaluation method, please refer to the following content.

[0049] The specific implementation method for segmenting multiple screenshots according to a preset segmentation path in step S120 to obtain the evaluation image group corresponding to each segmentation path can be as follows:

[0050] Step S121: Divide each screenshot into multiple first screenshot sub-regions and multiple second screenshot sub-regions based on preset viewfinder dividing lines, such as... Figure 3 As shown, the first and second screenshot sub-regions are interspersed in the vertical direction. The first screenshot sub-region represents the non-overlapping area of ​​lime spraying, and the second screenshot sub-region represents the overlapping area of ​​lime spraying.

[0051] Step S122: Classify multiple first screenshot sub-regions according to the preset segmentation path to obtain multiple first screenshot sub-regions corresponding to the non-overlapping area of ​​each nozzle, and arrange them sequentially according to the movement path of the nozzle to obtain the first evaluation image group corresponding to the non-overlapping area of ​​each nozzle. The first evaluation image group corresponds to the central area of ​​the fan-shaped spray of a nozzle.

[0052] Step S123: Classify multiple second screenshot sub-regions according to the preset segmentation path to obtain multiple second screenshot sub-regions corresponding to the overlapping areas of two adjacent nozzles, and arrange them sequentially according to the movement path of the nozzles to obtain the second evaluation image group corresponding to the overlapping areas of two adjacent nozzles. The second evaluation image group corresponds to the fan-shaped spray edge area of ​​two adjacent nozzles.

[0053] Step S124: The color-based image feature extraction algorithm sequentially evaluates the first lime spraying anomaly frequency index of the nozzle corresponding to each first evaluation image group;

[0054] Step S125: The color-based image feature extraction algorithm sequentially evaluates the second lime spraying abnormality frequency index and the third lime spraying abnormality frequency index of the two adjacent nozzles corresponding to each second evaluation image group.

[0055] Step S126: Calculate the first lime spraying abnormal frequency index, the second lime spraying abnormal frequency index, and the third lime spraying abnormal frequency index for each nozzle, and comprehensively evaluate the working condition index of the nozzle through a weighted algorithm. If the working condition index is less than the preset threshold, it is determined that the nozzle spraying is abnormal, and a corresponding abnormal warning instruction is generated so that the staff can adjust or replace the corresponding abnormal nozzle in time.

[0056] The specific calculation method for the first lime spraying abnormal frequency index corresponding to the non-overlapping area can be as follows:

[0057] Step S1241: Obtain the first screenshot sub-region in the first evaluation image group, divide it into multiple lime distribution evaluation areas, extract white pixels in each lime distribution evaluation area based on a color-based image recognition algorithm, calculate the proportion of white pixels in the total pixels of the lime distribution evaluation area, and then obtain the lime powder coverage ratio in each lime distribution evaluation area. Calculate the average coverage ratio of lime powder and the standard deviation of multiple coverage ratios for multiple lime distribution evaluation areas in the first screenshot sub-region.

[0058] Step S1242: Based on the average coverage ratio and standard deviation, a weighted algorithm is used to calculate the spraying quality index corresponding to the first screenshot sub-region. If the spraying quality index exceeds a threshold, the first screenshot sub-region is marked as the first abnormal spraying, and the number of first abnormal sprayings in the first evaluation map group is counted. Then, the first abnormal frequency of the nozzle corresponding to the first evaluation map is calculated and recorded as the first lime spraying abnormal frequency index. For example, the number of first abnormal sprayings divided by the camera shooting time can obtain the corresponding first abnormal frequency. The specific principle is roughly as follows: the calculation model has an ideal average coverage ratio and a reasonable standard deviation threshold. The greater the deviation, the larger the value of the first lime powder spraying abnormal index, that is, the greater the deviation between the detected state and the ideal spraying state. The average coverage ratio represents the overall coverage ratio, and the standard deviation is used to evaluate the uniformity of lime powder distribution.

[0059] The specific calculation methods for the second and third lime spraying abnormal frequency indices corresponding to the superimposed area can be as follows:

[0060] Step S1251: Obtain the second screenshot sub-region in the second evaluation image group, divide it into multiple lime distribution evaluation areas, extract the white pixels in each lime distribution evaluation area based on the color-based image recognition algorithm, calculate the proportion of white pixels in the total pixels of the lime distribution evaluation area, and then obtain the lime powder coverage ratio in each lime distribution evaluation area, and calculate the average lime powder coverage ratio corresponding to multiple lime distribution evaluation areas in the second screenshot sub-region.

[0061] Step S1252: If the average coverage ratio is 50%-70% of the ideal average coverage value, then retrieve the average coverage ratios of the two first screenshot sub-regions adjacent to the second screenshot sub-region, and add a second lime spraying anomaly to the nozzle corresponding to the first screenshot sub-region with the larger average coverage ratio.

[0062] Step S1253: If the average coverage ratio is 5%-49% of the ideal average coverage value, retrieve the average coverage ratios of the two first screenshot sub-regions adjacent to the second screenshot sub-region, and add a second lime spraying anomaly to the nozzle corresponding to the first screenshot sub-region with the larger average coverage ratio.

[0063] Furthermore, if the average coverage ratio is smaller than the ideal average coverage value, the nozzle corresponding to the first screenshot sub-area with the smaller average coverage ratio will be subject to a third lime spraying anomaly.

[0064] Step S1254: Count the number of times the second lime spraying abnormality and the third lime spraying abnormality occurred for each nozzle, and then obtain the second lime spraying abnormality frequency index and the third lime spraying abnormality frequency index corresponding to each nozzle.

[0065] A brief description of the processing logic of steps S1252-S1253 above: The abnormal causes of the second evaluation map group are mainly due to two nozzles. When an abnormal segment occurs, that is, when the average coverage ratio of lime powder in a certain second screenshot sub-region is 50%-70% of the ideal average coverage value, it is highly likely that one nozzle is abnormal. Even if the second nozzle is abnormal, it is a relatively minor abnormality. In this case, the abnormality is uniformly attributed to one nozzle abnormality. This abnormal nozzle is the nozzle corresponding to the first screenshot sub-region with a larger average coverage ratio. Because its first screenshot sub-region has more coverage, the amount sprayed in its corresponding second screenshot sub-region is naturally less. Therefore, the abnormality is attributed to this abnormal nozzle.

[0066] However, when the average coverage ratio of lime powder in a certain second screenshot sub-region is 5%-49% of the ideal average coverage value, this state is usually caused by two adjacent nozzles. However, it is necessary to distinguish between the primary and secondary causes. The method of distinction is still to compare the average coverage ratio values ​​of the two nozzles in the corresponding first screenshot sub-regions. The larger value is the primary cause, and the smaller value is the secondary cause. However, there is a special case: when the average coverage ratio value of the first screenshot sub-region of the secondary cause is greater than the ideal average coverage value, it will be attributed to the secondary cause. By counting the number of primary and secondary causes and combining the frequency of the abnormal number of the previous non-overlapping area, the working status of each nozzle is comprehensively determined by a weighted algorithm.

[0067] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for monitoring and controlling the ash-spreading status based on image recognition, the method being applicable to a silkworm rearing disinfection system, the system comprising a silkworm rearing disinfection surface arranged in a square strip shape, a spray frame, and a camera, wherein the spray frame and camera move along the long side of the silkworm rearing disinfection surface via a guide rail above the disinfection surface, wherein... The spraying frame is evenly equipped with multiple nozzles, which are used to spray lime powder from the storage hopper in a fan shape onto the disinfection surface of the silkworm rearing platform. The disinfection surface is green in color. The method is characterized by comprising: One minute after the ash-spraying operation is completed, the camera passes over the silkworm platform disinfection surface at a preset speed along the guide rail, continuously recording the status video of the silkworm platform disinfection surface and sending the status video to the cloud platform server. The cloud platform server can then evaluate the quality of the ash-spraying operation and determine the abnormal factors based on the status video of the silkworm platform disinfection surface in case of ash-spraying abnormalities. The ash-spraying operation is the operation of the spraying frame spraying lime powder onto the silkworm platform disinfection surface through the guide rail. Secondly, the PTZ server evaluates the quality of the ash-spreading operation, including: The guide rail running speed is obtained, and the corresponding screenshot interval is retrieved from the running parameter reference table based on the guide rail running speed. This results in multiple screenshots of the status video arranged in a time series, and the disinfection area of ​​the silkworm platform corresponding to each screenshot does not overlap. Multiple screenshots are segmented according to a preset segmentation path to obtain an evaluation image group corresponding to each segmentation path. The lime powder spraying anomaly index corresponding to the evaluation image group is analyzed, and the lime powder spraying quality of the nozzle is evaluated based on the lime powder spraying anomaly index. Secondly, the multiple screenshots are segmented according to a preset segmentation path, resulting in an evaluation image set corresponding to each segmentation path, including: Based on the preset viewfinder dividing line, each screenshot is divided into multiple first screenshot sub-regions and multiple second screenshot sub-regions, wherein the first screenshot sub-regions and the second screenshot sub-regions are interspersed. The first screenshot sub-regions represent the non-overlapping areas of lime spraying, and the second screenshot sub-regions represent the overlapping areas of lime spraying. Multiple first screenshot sub-regions are classified according to a preset segmentation path to obtain multiple first screenshot sub-regions corresponding to each nozzle non-overlapping area. These sub-regions are then arranged sequentially according to the nozzle's movement path to obtain a first evaluation image group corresponding to each nozzle non-overlapping area. Multiple second screenshot sub-regions are classified according to a preset segmentation path, thereby obtaining multiple second screenshot sub-regions corresponding to the overlapping areas of two adjacent nozzles, and arranged sequentially according to the movement path of the nozzles, thereby obtaining the second evaluation image group corresponding to the overlapping areas of two adjacent nozzles. Secondly, analyze and evaluate the lime powder spraying anomaly index corresponding to the map group, including: A color-based image feature extraction algorithm sequentially evaluates the first lime spraying anomaly frequency index of the nozzles corresponding to each first evaluation image group. The color-based image feature extraction algorithm sequentially evaluates the second lime spraying abnormality frequency index and the third lime spraying abnormality frequency index of two adjacent nozzles corresponding to each second evaluation image group. The first, second, and third lime spraying abnormal frequency indices for each nozzle are statistically analyzed. The operating condition index of the nozzle is comprehensively evaluated through a weighted algorithm. If the operating condition index is less than a preset threshold, it is determined that the nozzle spraying is abnormal, and a corresponding abnormal warning instruction is generated so that the staff can adjust or replace the corresponding abnormal nozzle in time. Secondly, the color-based image feature extraction algorithm sequentially evaluates the second and third lime spraying anomaly frequency indices of the two adjacent nozzles corresponding to each second evaluation image group, including: The color-based image feature extraction algorithm calculates the second and third lime spraying anomaly frequency indices of the two adjacent nozzles corresponding to each second evaluation image group, including: The second screenshot sub-region in the second evaluation image group is obtained and divided into multiple lime distribution evaluation areas. White pixels in each lime distribution evaluation area are extracted based on a color-based image recognition algorithm. The proportion of white pixels in the total pixels of the lime distribution evaluation area is calculated to obtain the lime powder coverage ratio in each lime distribution evaluation area. The average lime powder coverage ratio corresponding to multiple lime distribution evaluation areas in the first screenshot sub-region is also calculated. If the average coverage ratio is 50%-70% of the ideal average coverage value, then retrieve the average coverage ratios of the two first screenshot sub-regions adjacent to the second screenshot sub-region, and add a second lime spraying anomaly to the nozzle corresponding to the first screenshot sub-region with the larger average coverage ratio. If the average coverage ratio is 5%-49% of the ideal average coverage value, then retrieve the average coverage ratios of the two first screenshot sub-regions adjacent to the second screenshot sub-region, and add a second lime spraying anomaly to the nozzle corresponding to the first screenshot sub-region with the larger average coverage ratio. Furthermore, if the average coverage ratio is smaller than the ideal average coverage value, the nozzle corresponding to the first screenshot sub-area with the smaller average coverage ratio will have one more instance of abnormal lime spraying. The number of times the second lime spraying abnormality occurred and the number of times the third lime spraying abnormality occurred for each nozzle were counted, and then the frequency index of the second lime spraying abnormality and the frequency index of the third lime spraying abnormality for each nozzle were obtained.

2. The image recognition-based ash-spreading state monitoring and control method according to claim 1, characterized in that, The color-based image feature extraction algorithm sequentially evaluates the first lime spraying anomaly frequency index of the nozzle corresponding to each first evaluation image group, including: The first screenshot sub-region in the first evaluation image group is obtained and divided into multiple lime distribution evaluation regions. White pixels in each lime distribution evaluation region are extracted based on a color-based image recognition algorithm. The proportion of white pixels in the total pixels of the lime distribution evaluation region is calculated to obtain the lime powder coverage ratio in each lime distribution evaluation region. The average coverage ratio of lime powder and the standard deviation of multiple coverage ratios are calculated for the multiple lime distribution evaluation regions in the first screenshot sub-region. Based on the average coverage ratio and standard deviation, the spraying quality index corresponding to the first screenshot sub-region is calculated by a weighted algorithm. If the spraying quality index exceeds the threshold, the first screenshot sub-region is marked as the first abnormal spraying. The number of first abnormal sprayings in the first evaluation map group is counted, and then the first abnormal frequency of the nozzle corresponding to the first evaluation map is calculated, which is denoted as the first lime spraying abnormal frequency index.

Images