A rainwater outlet visual management method and system

Abstract

The application discloses a rain drain visual management method and system, the method comprises the following steps: acquiring original video frames and environmental parameters of a rain drain area, the environmental parameters include illumination intensity, weather type and night vision state; performing environmental correction on the original video frames based on the environmental parameters to obtain corrected video frames; extracting water color features, texture features and floating object area features from the corrected video frames, and performing multi-dimensional feature fusion to obtain comprehensive abnormal feature values; performing dynamic interference filtering on the comprehensive abnormal feature values based on the comparison result of a real-time dynamic change value and a preset dynamic threshold to obtain effective abnormal feature values; calculating an abnormal confidence level according to the effective abnormal feature values; and triggering an alarm prompt of a corresponding level based on the abnormal confidence level. The application realizes automatic and accurate identification and hierarchical early warning of abnormal discharge of a rain drain, and effectively improves the environmental risk response efficiency and the management refinement level.

Description

Technical Field

[0001] This invention relates to the field of environmental protection monitoring technology, and in particular to a visual management method and system for rainwater discharge outlets. Background Technology

[0002] As a critical point for industrial enterprises to discharge rainwater, storm drains directly affect the quality of surrounding water bodies. This is especially true in industries such as steel, chemicals, metallurgy, and agricultural and food processing, where initial rainwater runoff and mixed pipe networks can easily lead to abnormal discharges, posing a potential impact on the water environment. Traditional factory storm drain management mainly relies on regular manual inspections and simple video monitoring, which has significant technical shortcomings and management loopholes.

[0003] On the one hand, manual inspections are limited by frequency, time, and weather. The usual 1-2 inspections per day cannot cover critical periods such as nighttime, heavy rain, and dense fog. Abnormal emission incidents are often not discovered until several hours after they occur, missing the best time for handling. Moreover, manual judgment relies on experience, and the accuracy rate for identifying subtle anomalies such as low-concentration abnormally colored water bodies and trace amounts of oil is less than 60%, making it prone to misjudgment or omission. On the other hand, existing simple video surveillance only has a "viewing" function and lacks intelligent analysis capabilities, making it unable to automatically identify anomalies. Data transmission mostly uses a single network, which can easily lead to data interruption when fiber optic cables fail. The storage system lacks redundancy design, resulting in a high risk of video data loss. Furthermore, the monitoring data is not linked with systems such as water quality monitoring, GIS geographic information, and environmental protection work orders, forming "information silos" that cannot achieve anomaly verification, source tracing analysis, and closed-loop handling. This leads to low emergency response efficiency, with an average handling time of more than 50 minutes, which is insufficient to meet the refined risk prevention and control requirements of environmental management departments. Summary of the Invention

[0004] In order to overcome the above-mentioned shortcomings of the prior art, the purpose of this invention is to provide a visual management method and system for rainwater discharge outlets.

[0005] The technical solution adopted by this invention to solve its technical problem is: a visual management method for rainwater drainage outlets, comprising: Acquire raw video frames and environmental parameters of the rainwater drainage area, including light intensity, weather type, and night vision status; Based on the environmental parameters, environmental correction is performed on the original video frame to obtain the corrected video frame; Based on the corrected video frames, water color features, texture features, and floating object area features are extracted, and multi-dimensional feature fusion is performed to obtain a comprehensive anomaly feature value. Based on the comparison results between the real-time dynamic change and the preset dynamic threshold, the comprehensive abnormal feature value is dynamically filtered for interference to obtain an effective abnormal feature value. Calculate the anomaly confidence level based on the effective anomaly feature values; Based on the aforementioned abnormal confidence level, an alarm notification of the corresponding level is triggered.

[0006] As a further improvement of the present invention: the formula for environmental correction of the original video frame is: ; in, To correct the pixel value of the video frame at coordinates (x, y); This represents the pixel value of the original video frame at coordinates (x, y). This represents the standard value for normal monitoring illumination of the rainwater drainage outlet; Real-time light intensity; This is a weather correction factor; This is the infrared night vision compensation factor; This is the illumination weighting coefficient; This is the weather weighting coefficient.

[0007] As a further improvement of the present invention, the multi-dimensional feature fusion step is performed using the following formula: ; in, The comprehensive anomaly characteristic value of the rainwater drainage outlet; The real-time water color feature value is obtained by extracting the RGB mean of the corrected video frames; This is the baseline value for the color characteristics of normal rainwater; For oil / foam texture coefficient; The real-time area of ​​floating objects / debris is obtained through contour detection of corrected video frames; This refers to the total area of ​​the monitoring screen of the high-definition infrared network camera; These are the color feature weighting coefficients; These are the texture feature weight coefficients; This is the area feature weighting coefficient.

[0008] As a further improvement of the present invention, the dynamic interference filtering step is performed using the following formula: ,in, These are the effective abnormal feature values ​​after filtering out interference; The comprehensive anomaly characteristic value of the rainwater drainage outlet; The value is a real-time dynamic change, obtained by calculating the average pixel difference between two adjacent corrected video frames; This is the normal dynamic threshold. Interference type coefficient; The weighting coefficients are for interference effects.

[0009] As a further improvement of the present invention: the anomaly confidence level is calculated using the following formula: ,in, For abnormal confidence levels; These are the effective abnormal feature values ​​after filtering out interference; The threshold for abnormal features; The maximum abnormal feature value; The quality score of the corrected frame is calculated based on the sharpness and contrast of the corrected video frame.

[0010] As a further improvement of the present invention, it also includes: calculating the emergency response priority based on the anomaly confidence level, the anomaly diffusion rate, and the importance level of the rainwater outlet; and automatically generating and allocating disposal work orders based on the emergency response priority.

[0011] As a further improvement of the present invention, the emergency response priority is calculated using the following formula: ,in, Prioritize emergency response; For abnormal confidence levels; The abnormal diffusion rate is expressed in m² / min and is estimated from multiple corrected video frames. The importance level of rainwater drainage outlets is determined by their proximity to the external drainage network, with rainwater drainage outlets located near the plant area having a higher importance level than those located within the plant area. This is the diffusion rate weighting coefficient.

[0012] As a further improvement of the present invention, the alarm prompt includes three levels: no alarm is triggered when the anomaly confidence level is less than the first threshold, a first-level alarm is triggered when the anomaly confidence level is between the first threshold and the second threshold, and a second-level alarm is triggered when the anomaly confidence level is greater than or equal to the second threshold.

[0013] The present invention also provides a visual management system for rainwater drainage outlets, comprising: The acquisition module is used to acquire raw video frames and environmental parameters of the rainwater drainage area, including light intensity, weather type and night vision status. The intelligent processing module is used for: Based on the environmental parameters, environmental correction is performed on the original video frame to obtain the corrected video frame; Based on the corrected video frames, water color features, texture features, and floating object area features are extracted, and multi-dimensional feature fusion is performed to obtain a comprehensive anomaly feature value. Based on the comparison results between the real-time dynamic change and the preset dynamic threshold, the comprehensive abnormal feature value is dynamically filtered for interference to obtain an effective abnormal feature value. Calculate the anomaly confidence level based on the effective anomaly feature values; The alarm display module is used to display the corrected video frames and comprehensive abnormal feature values, and trigger alarm prompts of the corresponding level based on the abnormal confidence level. The emergency management module is used to call a preset emergency response priority formula to calculate the anomaly confidence level and preset parameters, output the emergency response priority, and automatically generate and allocate disposal work orders based on the emergency response priority.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention combines video frames, environmental correction, and multi-dimensional anomaly feature fusion to effectively overcome the interference of complex environments (such as changes in lighting, weather effects, and nighttime imaging) on ​​video monitoring results, thereby improving the accuracy of anomaly identification. Based on anomaly confidence levels, it implements tiered alarms, making alarm information more distinctive and enabling managers to quickly assess the urgency of events and prioritize high-risk anomalies, thus improving emergency response efficiency. It eliminates the need for continuous manual intervention, achieving continuous, real-time intelligent monitoring of storm drain discharge status and reducing manual inspection costs and the risk of missed detections. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the process of the present invention.

[0016] Figure 2 This is a schematic diagram illustrating the calculation of anomaly confidence based on video and environmental data according to the present invention.

[0017] Figure 3 This is a schematic diagram illustrating the alarm triggering and emergency management of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0019] In order to solve the technical problems in the prior art, the present invention will now be further described in conjunction with the accompanying drawings and embodiments: like Figures 1 to 3 As shown, this invention discloses a visual management method for rainwater drainage outlets, comprising: S1: Acquire the original video frames and environmental parameters of the rainwater drainage area, including light intensity, weather type and night vision status; S2: Perform environmental correction on the original video frame based on environmental parameters to obtain the corrected video frame; In some implementations, the formula for environmental correction of the original video frames is as follows: ; in, To correct the pixel value of the video frame at coordinates (x, y); This represents the pixel value of the original video frame at coordinates (x, y). This represents the standard value for normal monitoring illumination of the rainwater drainage outlet; Real-time light intensity; This is a weather correction factor; This is the infrared night vision compensation factor; This is the illumination weighting coefficient; This is the weather weighting coefficient.

[0020] S3: Extract water color features, texture features, and floating object area features from the corrected video frames, and perform multi-dimensional feature fusion to obtain comprehensive anomaly feature values; In some implementations, the multi-dimensional feature fusion step is performed using the following formula: ; in, The comprehensive anomaly characteristic value of the rainwater drainage outlet; The real-time water color feature value is obtained by extracting the RGB mean of the corrected video frames; This is the baseline value for the color characteristics of normal rainwater; For oil / foam texture coefficient; The real-time area of ​​floating objects / debris is obtained through contour detection of corrected video frames; This refers to the total area of ​​the monitoring screen of the high-definition infrared network camera; These are the color feature weighting coefficients; These are the texture feature weight coefficients; This is the area feature weighting coefficient.

[0021] S4: Based on the comparison results between the real-time dynamic change and the preset dynamic threshold, dynamic interference filtering is performed on the comprehensive abnormal feature value to obtain effective abnormal feature value. In some implementations, the dynamic interference filtering step is performed using the following formula: ,in, These are the effective abnormal feature values ​​after filtering out interference; The comprehensive anomaly characteristic value of the rainwater drainage outlet; The value is a real-time dynamic change, obtained by calculating the average pixel difference between two adjacent corrected video frames; This is the normal dynamic threshold. Interference type coefficient; The weighting coefficients are for interference effects.

[0022] S5: Calculate the anomaly confidence level based on the valid anomaly characteristic values; In some implementations, the anomaly confidence level is calculated using the following formula: ,in, For abnormal confidence levels; These are the effective abnormal feature values ​​after filtering out interference; The threshold for abnormal features; The maximum abnormal feature value; The quality score of the corrected frame is calculated based on the sharpness and contrast of the corrected video frame.

[0023] S6: Trigger alarm prompts of the corresponding level based on the abnormal confidence level.

[0024] Some implementations also include: calculating emergency response priorities based on anomaly confidence level, anomaly diffusion rate, and the importance level of rainwater outlets; and automatically generating and assigning disposal work orders based on emergency response priorities.

[0025] Furthermore, the emergency response priority is calculated using the following formula: ,in, Prioritize emergency response; For abnormal confidence levels; The abnormal diffusion rate is expressed in m² / min and is estimated from multiple corrected video frames. The importance level of rainwater drainage outlets is determined by their proximity to the external drainage network, with rainwater drainage outlets located near the plant area having a higher importance level than those located within the plant area. This is the diffusion rate weighting coefficient.

[0026] Furthermore, the alarm notification includes three levels: no alarm is triggered when the anomaly confidence level is less than the first threshold, a first-level alarm is triggered when the anomaly confidence level is between the first and second thresholds, and a second-level alarm is triggered when the anomaly confidence level is greater than or equal to the second threshold.

[0027] This invention also discloses a visual management system for storm drains. By deploying high-definition infrared cameras and environmental sensors, the system collects real-time video and environmental data from the drain outlets. The system uses intelligent algorithms to perform environmental correction, multi-dimensional anomaly feature fusion, and dynamic interference filtering on the video, accurately identifying problems such as abnormal water color and floating debris accumulation. These issues are then visually displayed through heat maps. Combined with fiber optic + 4G / 5G redundant transmission, the system ensures real-time data transmission. The emergency management module automatically allocates work orders based on anomaly priority, improving emergency response efficiency by more than 40% and significantly reducing the frequency of misjudgments and invalid inspections, thus achieving intelligent and precise management of storm drains.

[0028] The acquisition module includes a high-definition infrared network camera and an environmental auxiliary sensor. The high-definition infrared network camera acquires the pixel values ​​of the original video frames in the rain drain area, while the environmental auxiliary sensor acquires the real-time ambient light intensity, weather type, and infrared night vision trigger status, and records the total area of ​​the camera's monitoring screen. Furthermore, in the acquisition module, the high-definition infrared network camera has a resolution of no less than 4 million pixels, an infrared night vision distance of no less than 50 meters, and its installation location is determined through 3D modeling to ensure coverage of the rainwater drainage area and the passage within a 5-meter radius around it; the environmental auxiliary sensor has a sampling frequency of 1Hz, a light intensity measurement range of 0~100,000 lux, and a weather type recognition accuracy of no less than 95%, including four weather types: rain, fog, snow, and sunny.

[0029] Transmission and storage module: includes fiber optic transmission link, dedicated storage server and formula reference parameter library, raw data output by the link transmission acquisition module, video data stored on the server, and normal illumination standard value, normal rainwater color reference value and normal dynamic threshold value of rainwater discharge outlet in the parameter library. Furthermore, in the transmission and storage module, the fiber optic transmission link has a transmission rate of no less than 100Mbps and is equipped with a 4G / 5G wireless redundant link, which automatically switches within 10 seconds in the event of a fiber optic failure; the dedicated storage server uses H.265 encoding to compress video data, is configured with a 24-bay RAID5 array, and has a total storage capacity of no less than 16TB; the formula benchmark parameter library supports automatic updates on a quarterly basis, with the updated data coming from normal rainwater drainage outlet monitoring samples from the past 90 days, which are monitoring data that have been manually verified to be free of abnormalities.

[0030] Intelligent processing module: It calls the preset video frame environment correction formula, multi-dimensional anomaly feature fusion formula, dynamic interference filtering formula and anomaly confidence calculation formula to calculate the video data and benchmark parameters output by the transmission and storage module, and outputs the corrected video frame pixel value, comprehensive anomaly feature value, effective anomaly feature value and anomaly confidence in sequence.

[0031] Furthermore, in the intelligent processing module, the video frame environment correction formula is: ,in, To correct the pixel value of the video frame at coordinates (x, y); This represents the pixel value of the original video frame at coordinates (x, y). This represents the standard value for normal monitoring illumination of the rainwater drainage outlet; Real-time light intensity; This is a weather correction factor, corresponding to rain, fog, snow, and sunny days. The values ​​are 0.85, 0.7, 0.65, and 1.0, respectively. This is an infrared night vision compensation factor, selectable for nighttime and daytime use. They are 1.2 and 1.0 respectively; This is the illumination weighting coefficient, which is optional. =0.4; Weather weighting coefficient, optional =0.3.

[0032] Furthermore, in the intelligent processing module, the formula for fusing multi-dimensional anomaly features is: ,in, The comprehensive anomaly characteristic value of the rainwater drainage outlet; The real-time water color feature value is obtained by extracting the RGB mean of the corrected video frames; This is the baseline value for the color characteristics of normal rainwater; This is the oil / foam texture coefficient, selectable for normal rainwater and oil stains. The values ​​are 0.1 and 0.7~0.9 respectively; The real-time area of ​​floating objects / debris is obtained through contour detection of corrected video frames; This refers to the total area of ​​the monitoring screen of the high-definition infrared network camera; These are color feature weighting coefficients, which are optional. =0.4; These are texture feature weight coefficients, which are optional. =0.5; These are area feature weighting coefficients, which are optional. =0.3; This formula is used to integrate three types of abnormal features: color, texture, and area, to achieve a unified characterization of multiple types of pollution.

[0033] Furthermore, in the intelligent processing module, the dynamic interference filtering formula is: ,in, These are the effective abnormal feature values ​​after filtering out interference; The comprehensive anomaly characteristic value of the rainwater drainage outlet; The value is a real-time dynamic change, obtained by calculating the average pixel difference between two adjacent corrected video frames; This is the normal dynamic threshold. For interference type coefficients, selectable options include rain swaying, leaf swaying, and no interference. The values ​​are 0.6, 0.8, and 0.1 respectively. To mitigate the impact of interference on weighting coefficients, optional =0.7; This formula is used to filter out natural dynamic interference and reduce false alarms caused by rain and leaf movement.

[0034] Furthermore, in the intelligent processing module, the formula for calculating the anomaly confidence level is: ,in, The anomaly confidence level has a value range of [0, 1]. These are the effective abnormal feature values ​​after filtering out interference; For abnormal feature thresholds, optional =0.2; The maximum abnormal feature value, optional =1.5; The quality score of the corrected frame is calculated based on the sharpness and contrast of the corrected video frame, with a value range of [0.6, 1.0].

[0035] Furthermore, in the alarm display module, the multi-screen monitoring display supports 4 / 8 / 16-screen splitting, and the client has video playback function and PTZ control interface; the distribution of comprehensive abnormal feature values ​​is displayed using a heat map, with red indicating high-anomaly areas and blue indicating normal areas; the warning threshold is divided into three levels. <0.5 will not trigger an alarm, 0.5≤ A value less than 0.8 triggers a yellow alert. A value ≥0.8 triggers a red alarm, and the alarm information includes the location of the anomaly, the type of anomaly, and the confidence level of the anomaly.

[0036] Furthermore, in the emergency management module, the formula for emergency response priority is: ,in, This represents the priority of emergency response, with a value range of [0, 1]. For abnormal confidence levels; The abnormal diffusion rate is expressed in m² / min and is estimated from multiple corrected video frames. The importance level of the rainwater discharge outlet is defined by a value ranging from [0.5 to 1.0]. Rainwater discharge outlets located near the external drainage network have a higher importance level than those located inside the plant area. This is a diffusion rate weighting coefficient, which can be selected. =0.6.

[0037] Alarm Display Module: Configures a multi-screen monitoring screen and client to display the corrected video frames and comprehensive abnormal feature value distribution output by the intelligent processing module, and sets early warning thresholds based on the abnormal confidence level to trigger corresponding alarm prompts; Emergency Management Module: It calls the preset emergency response priority formula to calculate the anomaly confidence level and preset parameters, outputs the emergency response priority, allocates handling work orders accordingly, and records the calculation data of the whole process to form an anomaly handling ledger.

[0038] The system of this invention deploys high-definition infrared network cameras and environmental auxiliary sensors, combined with innovative algorithms in the intelligent processing module such as multi-dimensional anomaly feature fusion, dynamic interference filtering, and anomaly confidence calculation, to accurately monitor the drainage status of rainwater outlets. The system can automatically identify key features such as abnormal water color and floating debris accumulation, and intuitively display the distribution of anomalies through heat maps. With a three-level early warning mechanism, it ensures that management personnel can respond quickly to potential risks. At the same time, the fiber optic + 4G / 5G redundant transmission design ensures data real-time performance. This architecture improves emergency response efficiency by more than 40% compared to traditional manual inspections, and significantly reduces the missed alarm rate and false alarm rate.

[0039] By using a video frame environment correction formula to compensate for lighting, weather, and night vision conditions in real time, and combining it with a multi-dimensional anomaly feature fusion algorithm, the problem of misjudgment caused by the distortion of monitoring images in complex environments is effectively solved. At the same time, the emergency management module uses an emergency response priority formula to quantify the anomaly confidence, spread speed, and importance level of rainwater outlets, automatically generating priority handling work orders and forming a full-process ledger. This design greatly improves the efficiency of operation and maintenance resource allocation and can significantly reduce the frequency of invalid inspections.

[0040] Implementation Case 1: This invention discloses a visual management system for rainwater drainage outlets, applied to the management of rainwater drainage outlets in a large metallurgical enterprise. The enterprise has five rainwater drainage outlets located in key areas such as the production unit area, storage tank area, and loading / unloading area. Outlet No. 3 is adjacent to a river outside the plant, requiring special attention to prevent the risk of oil spills and chemical residues being discharged with rainwater. The enterprise uses this system to conduct full-process management of the rainwater drainage outlets.

[0041] Data Acquisition Module Deployment: High-definition infrared network cameras were installed at five drainage outlets, with camera resolution adjusted to 4 megapixels and infrared night vision distance set to 55 meters. Installation positions and angles were determined through 3D modeling: cameras at drainage outlets 1 and 2 in the production unit area were installed at a height of 3.2 meters, primarily covering the drainage outlet and the surrounding 3-meter-wide pipe network interface; camera at drainage outlet 3 in the tank area was installed at a height of 3.5 meters, simultaneously covering the drainage outlet, the riverbank, and the surrounding 5-meter-wide passageway; cameras at drainage outlets 4 and 5 in the loading and unloading area were installed at a height of 3 meters, focusing on the drainage outlet and the edge of the loading and unloading area. Each camera was equipped with an environmental auxiliary sensor, with a sampling frequency of 1Hz, to collect real-time data on the light intensity, weather type (e.g., rain, fog, sunny), and infrared night vision trigger status around each drainage outlet, and to record the total monitoring area of ​​each camera (1920×1080 pixels), ensuring no blind spots and complete data acquisition.

[0042] Transmission and storage module setup: Single-mode fiber optic cables are routed from the acquisition devices at five drainage outlets and connected to the fiber optic switches of the company's existing integrated security platform. The transmission rate was stabilized at 120Mbps to ensure real-time transmission of video data and environmental parameters. A 4G wireless module was added to the transmission link at each drainage outlet as redundancy. Through multiple tests and adjustments to the switching mechanism, it was ensured that in the event of a fiber optic link failure, automatic switching to the 4G link would occur within 10 seconds to avoid data interruption. A dedicated storage server was deployed in the company's central computer room, configured with a 24-bay RAID5 disk array, providing a total storage capacity of 20TB. H.265 encoding was used to compress video data, reducing storage space usage. Simultaneously, a formula-based benchmark parameter library was established, recording standard values ​​for normal monitoring illumination, normal rainwater color, and normal dynamic thresholds for drainage outlets. The parameter library is set to be automatically updated quarterly, with updated data sourced from drainage outlet monitoring samples that have been manually verified frame-by-frame for anomalies within the past 90 days, ensuring that the benchmark parameters accurately reflect the actual scenario.

[0043] Intelligent processing module operation: An intelligent processing module is deployed on the server in the central computer room, and the module is set to process the acquired video data and environmental parameters every 30 seconds. First, the video frame environment correction formula is called. The video frame environment correction formula is: ,in, To correct the pixel value of the video frame at coordinates (x, y); This represents the pixel value of the original video frame at coordinates (x, y). This represents the standard value for normal monitoring illumination of the rainwater drainage outlet; Real-time light intensity; This is a weather correction factor, corresponding to rain, fog, snow, and sunny days. The values ​​are 0.85, 0.7, 0.65, and 1.0, respectively. This refers to the infrared night vision compensation factor, corresponding to nighttime and daytime conditions. They are 1.2 and 1.0 respectively; The illumination weighting coefficient and =0.4; Weather weighting coefficient and =0.3, combined with real-time light intensity, weather type, and infrared night vision trigger status collected by the sensor, pixel-level correction is performed on the original video frames to eliminate the impact of environmental factors such as strong light, rain reflection, and fog blur on image quality; then, the multi-dimensional anomaly feature fusion formula is called to extract features such as water color, oil stain texture, and floating object area from the corrected video frames, and the comprehensive anomaly feature value is obtained by fusion calculation. The multi-dimensional anomaly feature fusion formula is: ,in, The comprehensive anomaly characteristic value of the rainwater drainage outlet; The real-time water color feature value is obtained by extracting the RGB mean of the corrected video frames; This is the baseline value for the color characteristics of normal rainwater; This represents the oil / foam texture coefficient, corresponding to normal rainwater and oil stains. The values ​​are 0.1 and 0.7~0.9 respectively; The real-time area of ​​floating objects / debris is obtained through contour detection of corrected video frames; This refers to the total area of ​​the monitoring screen of the high-definition infrared network camera; Here are the color feature weight coefficients, and =0.4; These are the texture feature weight coefficients, and =0.5; Here are the area feature weighting coefficients, and =0.3; Then, the dynamic interference filtering formula is called to compare the real-time dynamic changes with the normal dynamic threshold, filtering out natural interference signals such as rainwater flow, leaf swaying, and light and shadow changes, while retaining effective abnormal features. The dynamic interference filtering formula is: ,in, These are the effective abnormal feature values ​​after filtering out interference; The comprehensive anomaly characteristic value of the rainwater drainage outlet; The value is a real-time dynamic change, obtained by calculating the average pixel difference between two adjacent corrected video frames; This is the normal dynamic threshold. These are the interference type coefficients, corresponding to rain swaying, leaf swaying, and no interference. The values ​​are 0.6, 0.8, and 0.1 respectively. To influence the weighting coefficients and =0.7; Finally, the anomaly confidence calculation formula is called. The anomaly confidence calculation formula is: ,in, The anomaly confidence level has a value range of [0, 1]. These are the effective abnormal feature values ​​after filtering out interference; The threshold for abnormal features and =0.2; The largest abnormal feature value and =1.5; To calculate the quality score of the corrected frame, a value range of [0.6, 1.0] is used, based on the sharpness and contrast of the corrected video frame. Combining this with the corrected frame quality score, an anomaly confidence level is output for each rainwater drain outlet. If the confidence level is ≥0.5, it is marked as a potential anomaly and proceeds to the subsequent alarm process. Figure 2 As shown.

[0044] Alarm Display Module Application: Multi-screen monitoring screens are deployed in the company's environmental management office and production command center, supporting 16-screen splitting and simultaneously displaying real-time video frames from 5 storm drain outlets after correction. Clients are configured for environmental management personnel and production schedulers, supporting video playback (0.5-16x adjustable speed) and remote PTZ control, allowing for real-time adjustment of camera focus and angle to view detailed footage. The large screens display the distribution of comprehensive anomaly characteristic values ​​in a heat map format, with red areas representing high anomaly risk and blue areas representing normal conditions, providing a clear visual representation of each storm drain outlet's status. When the intelligent processing module outputs an anomaly confidence level, the module triggers an alarm based on preset thresholds: a yellow alarm is triggered when the confidence level is between 0.5 and 0.8, and a red alarm is triggered when the confidence level is ≥0.8. Alarm information is simultaneously displayed on the monitoring screens and pushed to the corresponding management personnel via the client. The information clearly indicates the anomaly storm drain outlet number, specific location, anomaly type, and anomaly confidence level, facilitating rapid problem location.

[0045] Emergency Management Module Handling: Upon receiving an alarm message, the emergency management module automatically invokes the emergency response priority formula, which is as follows: ,in, This represents the priority of emergency response, with a value range of [0, 1]. For abnormal confidence levels; The abnormal diffusion rate is expressed in m² / min and is estimated from multiple corrected video frames. The importance level of the storm drain is determined by a value ranging from [0.5 to 1.0]. Storm drains near municipal pipe networks have a higher importance level than storm drains within the factory area. The diffusion rate weighting coefficient and =0.6, combined with the anomaly confidence level, anomaly propagation speed (estimated based on continuous multi-frame video), and the importance level of the storm drain (outlet No. 3 is set to 1.0 due to its proximity to the river, and the other storm drains are set to 0.7), the emergency response priority is calculated. Work orders are assigned according to priority: when priority < 0.5, a basic work order is generated and assigned to frontline maintenance personnel, requiring on-site verification and handling within 4 hours; when priority 0.5-0.8, an emergency work order is generated and assigned to the environmental protection team leader, requiring handling within 2 hours; when priority ≥ 0.8, a top-priority work order is generated and assigned to the company's environmental protection manager, requiring handling within 1 hour. After receiving the work order through the client, the responsible person goes to the site to verify and handle the situation, uploading on-site photos, videos, and text records during the process; after handling, the system automatically verifies the results and marks the work order as closed-loop. Simultaneously, the system records the entire process calculation data from data collection to work order closure, forming an anomaly handling ledger. The ledger can be exported as a standardized report for internal management review and external compliance inspections.

[0046] Implementation Case 2: This invention discloses a visual management system for rainwater drainage outlets, applied to the management of rainwater drainage outlets in a medium-sized food processing enterprise. This enterprise mainly produces baked goods and beverages. There are two rainwater drainage outlets within the factory area: Outlet 1 is located outside the raw material warehouse, collecting rainwater from the warehouse roof and surrounding area; Outlet 2 is located next to the cleaning workshop, collecting rainwater from the area outside the workshop and the equipment cleaning area. It is necessary to prevent food residue and detergent residue from being discharged with the rainwater. The enterprise implements this system for rainwater drainage outlet management as follows: Data Acquisition Module Deployment: High-definition infrared network cameras were installed at drainage outlets No. 1 and No. 2, with a resolution of 4 megapixels and an infrared night vision distance of 50 meters. Installation locations were determined through 3D modeling: Camera No. 1 was installed at a height of 2.8 meters, with its lens facing the drainage outlet and the outer wall of the raw material warehouse, covering a 5-meter radius to prevent raw material residue from being washed into the drainage outlet by rainwater; Camera No. 2 was installed at a height of 3 meters, with its lens covering the drainage outlet, the drainage pipe outlet on the outer wall of the cleaning workshop, and a 4-meter radius around it, focusing on monitoring detergent residue. An environmental auxiliary sensor was installed next to each camera, with a sampling frequency of 1Hz, collecting data on light intensity, weather type (rain, snow, sunny), and infrared night vision trigger status, recording the total monitored area of ​​both cameras (1920×1080 pixels each), ensuring no blind spots in key areas.

[0047] Transmission and storage module setup: Fiber optic transmission links connect the acquisition devices at two drainage outlets to the company's central control room, with the transmission rate debugged to 100Mbps to ensure real-time and stable transmission of video data and environmental parameters. A 5G wireless redundant link is configured for each drainage outlet, and the switching logic is optimized through on-site testing to ensure that in the event of a fiber optic link failure, a switch to a 5G link can be made within 8 seconds to avoid data loss. A dedicated storage server is deployed in the central control room, configured with a 24-bay RAID5 disk array, with a total storage capacity of 18TB. H.265 encoding is used to compress video data, meeting the video storage needs for more than 6 months. A formula-based benchmark parameter library is established, recording the company's normal illumination standard values, normal rainwater color benchmark values, and normal dynamic thresholds for the drainage outlets. The parameter library is set to be automatically updated quarterly, with updated data derived from monitoring samples that have been jointly verified by workshop management and environmental protection specialists within the past 90 days and show no abnormalities, ensuring that the benchmark parameters conform to the company's actual production scenario.

[0048] Intelligent processing module operation: An intelligent processing module is deployed on the server in the central control room, and the module is set to process the collected data once every minute. First, the video frame environment correction formula is called, and combined with the light, weather and night vision status data collected by the sensors, the original video frames are corrected to solve problems such as dim night scenes and reflections in rainy weather, and improve the image clarity. Next, the multi-dimensional anomaly feature fusion formula is called to extract features such as water color (to determine whether there is abnormal color caused by detergent) and floating object area (to determine whether there is food residue) from the corrected video frames, and fuse them to calculate a comprehensive anomaly feature value. Then, the dynamic interference filtering formula is called to filter interference signals such as rainwater flow, wind-blown leaves, and changes in the shadow of workshop equipment, and retain the true anomaly features. Finally, the anomaly confidence calculation formula is called, and combined with the quality score of the corrected frame, the anomaly confidence of the two rainwater outlets is output. When the confidence is ≥0.5, the subsequent alarm is triggered.

[0049] Alarm Display Module Application: A multi-screen monitoring display is deployed in the enterprise's central control room, supporting 8-screen splitting. It displays real-time corrected video frames from drainage outlets 1 and 2, using heatmaps to annotate the distribution of comprehensive abnormal characteristic values. Red areas indicate abnormal risks, while blue areas indicate normal conditions. Clients are configured for workshop directors and environmental protection specialists. These clients support video playback and PTZ control, allowing for the retrieval of historical videos at any time, or remote adjustment of camera angles and focal lengths to observe drainage outlet details. When the anomaly confidence level reaches 0.5-0.8, a yellow alarm is displayed on the large screen, and the client pushes an alarm message. When the confidence level is ≥0.8, a red alarm is displayed on the large screen, simultaneously triggering an audio-visual alert in the central control room. The alarm message clearly states the abnormal drainage outlet number, anomaly type (e.g., "floating object anomaly," "color anomaly"), and confidence level, facilitating quick situation assessment for management personnel.

[0050] Emergency Management Module Handling: Upon receiving an alarm, the emergency management module invokes the emergency response priority formula, combining the anomaly confidence level, anomaly diffusion speed (e.g., the diffusion speed of floating food residue), and the importance level of the storm drain (0.9 for storm drain No. 2 due to its proximity to the cleaning workshop, and 0.6 for storm drain No. 1) to calculate the priority. Work orders are assigned based on priority: Priority < 0.5 is assigned to workshop maintenance personnel, requiring on-site response within 2 hours; priority 0.5-0.8 is assigned to environmental specialists, requiring response within 1 hour; priority ≥ 0.8 is assigned to the production supervisor, requiring response within 30 minutes. Upon arrival, the responsible personnel verify the cause of the anomaly and take appropriate measures (e.g., cleaning up food residue, sealing detergent leaks). During this process, on-site photos and response records are uploaded to the client. After completion, the system verifies the results and marks the work order as closed-loop. The system automatically records all process data, creating an anomaly response ledger and generating a monthly storm drain management report. This report is used to optimize workshop cleaning processes, adjust the frequency of routine storm drain maintenance, and improve the efficiency of enterprise storm drain management. Figure 3As shown.

[0051] In summary, after reading this invention document, those skilled in the art can make various other corresponding modifications to the technical solutions and concepts based on this invention without creative mental effort, and all of these modifications fall within the scope of protection of this invention.

Claims

1. A visual management method for rainwater drainage outlets, characterized in that, include: Acquire raw video frames and environmental parameters of the rainwater drainage area, including light intensity, weather type, and night vision status; Based on the environmental parameters, environmental correction is performed on the original video frame to obtain the corrected video frame; Based on the corrected video frames, water color features, texture features, and floating object area features are extracted, and multi-dimensional feature fusion is performed to obtain a comprehensive anomaly feature value. Based on the comparison results between the real-time dynamic change and the preset dynamic threshold, the comprehensive abnormal feature value is dynamically filtered for interference to obtain an effective abnormal feature value. Calculate the anomaly confidence level based on the effective anomaly feature values; Based on the aforementioned abnormal confidence level, an alarm notification of the corresponding level is triggered.

2. The method for visual management of rainwater drainage outlets according to claim 1, characterized in that, The formula for environmental correction of the original video frame is: ； in, To correct the pixel value of the video frame at coordinates (x, y); This represents the pixel value of the original video frame at coordinates (x, y). This represents the standard value for normal monitoring illumination of the rainwater drainage outlet; Real-time light intensity; This is a weather correction factor; This is the infrared night vision compensation factor; This is the illumination weighting coefficient; This is the weather weighting coefficient.

3. The method for visual management of rainwater drainage outlets according to claim 1, characterized in that, The multi-dimensional feature fusion step is performed using the following formula: ； in, The comprehensive anomaly characteristic value of the rainwater drainage outlet; The real-time water color feature value is obtained by extracting the RGB mean of the corrected video frames; This is the baseline value for the color characteristics of normal rainwater; For oil / foam texture coefficient; The real-time area of ​​floating objects / debris is obtained through contour detection of corrected video frames; This refers to the total area of ​​the monitoring screen of the high-definition infrared network camera; These are the color feature weighting coefficients; These are the texture feature weight coefficients; This is the area feature weighting coefficient.

4. The method for visual management of rainwater drainage outlets according to claim 3, characterized in that, The dynamic interference filtering steps are performed using the following formula: ； in, These are the effective abnormal feature values ​​after filtering out interference; The comprehensive anomaly characteristic value of the rainwater drainage outlet; The value is a real-time dynamic change, obtained by calculating the average pixel difference between two adjacent corrected video frames; This is the normal dynamic threshold. Interference type coefficient; The weighting coefficients are for interference effects.

5. The method for visual management of rainwater drainage outlets according to claim 4, characterized in that, The anomaly confidence level is calculated using the following formula: ； in, For abnormal confidence levels; These are the effective abnormal feature values ​​after filtering out interference; The threshold for abnormal features; The maximum abnormal feature value; The quality score of the corrected frame is calculated based on the sharpness and contrast of the corrected video frame.

6. The method for visual management of rainwater drainage outlets according to claim 1, characterized in that, Also includes: The emergency response priority is calculated based on the aforementioned anomaly confidence level, anomaly diffusion rate, and importance level of the storm drain outlet; Emergency response priorities are used to automatically generate and assign work orders.

7. The method for visual management of rainwater drainage outlets according to claim 6, characterized in that, The emergency response priority is calculated using the following formula: ， in, Prioritize emergency response; For abnormal confidence levels; The abnormal diffusion rate is expressed in m² / min and is estimated from multiple corrected video frames. The importance level of rainwater drainage outlets is determined by their proximity to the external drainage network, with rainwater drainage outlets located near the plant area having a higher importance level than those located within the plant area. This is the diffusion rate weighting coefficient.

8. The method for visual management of rainwater drainage outlets according to claim 1, characterized in that, The alarm notification includes three levels: No alarm is triggered when the anomaly confidence level is less than the first threshold. A first-level alarm is triggered when the anomaly confidence level is between the first and second thresholds. A second-level alarm is triggered when the anomaly confidence level is greater than or equal to the second threshold.

9. A visual management system for rainwater drainage outlets, characterized in that, include: The acquisition module is used to acquire raw video frames and environmental parameters of the rainwater drainage area, including light intensity, weather type and night vision status. The intelligent processing module is used for: Based on the environmental parameters, environmental correction is performed on the original video frame to obtain the corrected video frame; Based on the corrected video frames, water color features, texture features, and floating object area features are extracted, and multi-dimensional feature fusion is performed to obtain a comprehensive anomaly feature value. Based on the comparison results between the real-time dynamic change and the preset dynamic threshold, the comprehensive abnormal feature value is dynamically filtered for interference to obtain an effective abnormal feature value. Calculate the anomaly confidence level based on the effective anomaly feature values; The alarm display module is used to display the corrected video frames and comprehensive abnormal feature values, and trigger alarm prompts of the corresponding level based on the abnormal confidence level.

10. A visual management system for rainwater drainage outlets according to claim 9, characterized in that, It also includes an emergency management module, which is used to call a preset emergency response priority formula to calculate the anomaly confidence level and preset parameters, output the emergency response priority, and automatically generate and allocate disposal work orders based on the emergency response priority.

Images