Search

CN-121982540-A - Intelligent recycling environment-friendly monitoring system and method based on computer vision

CN121982540ACN 121982540 ACN121982540 ACN 121982540ACN-121982540-A

Abstract

The invention discloses an intelligent recycling environment-friendly monitoring system and method based on computer vision, and belongs to the technical field of computer vision technology and environment-friendly monitoring. The system comprises an image acquisition module, an edge calculation module, an environment-friendly evaluation module, a data storage module and a display interaction module. The method is characterized in that the cleanliness of recycled materials is quantified through HSV color space analysis (pollution areas are identified), the integrity is evaluated through an edge detection algorithm (breakage is detected), visual quality characteristics are converted into renewable utilization parameters (formula: renewable utilization= (cleanliness score/100) × (integrity score/100) × (material base renewable rate)), and carbon emission reduction calculation coefficients are dynamically corrected according to the renewable utilization parameters (formula: carbon emission reduction = weight×carbon emission reduction coefficients×renewable utilization), so that a quantification bridge from visual characteristics to environment-friendly parameters is established. The method solves the problem that the traditional fixed coefficient method cannot reflect the distortion of the carbon data caused by the real quality of the recycled material (error + -50% of the traditional method), realizes accurate carbon footprint calculation based on the actual quality state, improves the accuracy of the carbon footprint by 70%, and reaches the error level of + -15%. The system also realizes the functions of 7×24 hours of environmental protection monitoring, blockchain tracing, abnormal early warning and the like, improves the monitoring coverage rate by 233%, reduces the cost by 65%, and is suitable for various scenes such as community recycling stations, recycling enterprises, environmental protection supervision and the like.

Inventors

  • YE XUANJUN
  • ZHUANG XIANGYOU
  • YE XUANWU
  • Peng Niwa
  • Ye Renzong

Assignees

  • 鲁班长（深圳）科技有限公司

Dates

Publication Date
20260505
Application Date
20260128

Claims (20)

  1. 1. An intelligent recycling environmental protection monitoring system based on computer vision, which is characterized by comprising: The image acquisition module is arranged at a key position of the recycling station and is used for acquiring the images of the recycled objects and the station environment images in real time; The edge calculation module is in communication connection with the image acquisition module, adopts a deep learning target detection algorithm to analyze the acquired images in real time, identifies the types and the quantity of the recovered matters, and evaluates the quality state of the recovered matters by adopting color distribution analysis to evaluate the cleanliness of the recovered matters, namely converting the image of the recovered matters into an HSV color space, counting the pixel duty ratio of the saturation (S) exceeding a preset threshold value to identify a polluted area, and generating a cleanliness score; the integrity of the recovery is evaluated through surface texture analysis, namely, the contour of the recovery is extracted through an edge detection algorithm, the continuity and regularity of the contour are calculated, breakage, notch and crack are detected, and an integrity score is generated; Calculating the renewable utilization rate according to a formula of 'renewable utilization rate= (cleanliness score/100) × (integrity score/100) ×material base renewable rate', wherein the cleanliness score and the integrity score are normalized to a range of 0.0-1.0 by adopting a score standard of 0-100, and participate in calculation, and the material base renewable rate is obtained by inquiring from a preset database according to the material type of the recycled material; The environment-friendly evaluation module is in communication connection with the edge calculation module and comprises a carbon footprint calculation unit, wherein the carbon footprint calculation unit maintains a material carbon emission coefficient database, stores raw production carbon emission and recycling carbon emission of various materials, calculates carbon emission reduction according to the formula of 'carbon emission reduction=recycle weight× (raw production carbon emission coefficient-recycling carbon emission coefficient) ×renewable utilization rate', wherein the renewable utilization rate is calculated by the edge calculation module based on visual feature analysis, and the recycle weight is obtained through image analysis and density parameter estimation or is obtained through direct measurement of an integrated weighing sensor; the data storage module is in communication connection with the edge calculation module and the environment-friendly evaluation module and is used for storing image data, identification results and environment-friendly evaluation data; The display interaction module is in communication connection with the environment-friendly evaluation module and is used for displaying environment-friendly benefits to users in real time and providing environment-friendly integral feedback; the modules work cooperatively, and accurate environment-friendly monitoring and evaluation based on the actual quality state are realized by converting the visual quality characteristics of the recycled materials into renewable utilization rate parameters and dynamically adjusting carbon emission reduction calculation coefficients.
  2. 2. The intelligent recycling environmental monitoring system of claim 1, wherein the image acquisition module comprises: The throwing monitoring camera is arranged above the throwing port of the recycling equipment, adopts a overlooking shooting angle and is used for shooting recycled objects thrown by a user; The site monitoring camera is arranged at the top or at the side of the recovery site and is used for monitoring the overall environmental condition of the site for 7X 24 hours; The resolution ratio of the putting monitoring camera and the station monitoring camera is not lower than 1080P, the frame rate is not lower than 30fps, and the infrared night vision function is supported.
  3. 3. The intelligent recycling environmental protection monitoring system according to claim 2, wherein the throwing monitoring camera is provided with an automatic light supplementing system, and the light supplementing intensity can be automatically adjusted according to the environmental illumination conditions, so that the recycled objects can be clearly shot under different illumination conditions.
  4. 4. The intelligent recycling environmental protection monitoring system according to claim 2, wherein the site monitoring camera supports wide-angle shooting and PTZ (pan-tilt-zoom) control, can cover the whole area of the site, and realizes the omnibearing monitoring of an unsupervised blind area.
  5. 5. The intelligent recycling environmental monitoring system of claim 1, wherein the edge calculation module comprises: the image preprocessing unit is used for carrying out preprocessing operations such as denoising, enhancing, correcting and the like on the acquired image; and the AI processing unit is used for analyzing the preprocessed image in real time by adopting a deep learning target detection algorithm based on YOLO or Faster R-CNN.
  6. 6. The intelligent recycling environmental monitoring system of claim 5, wherein the AI processing unit is capable of identifying more than 50 common recycle types including, but not limited to, PET plastic bottles, pop cans, cartons, glass bottles, waste paper, metal cans, plastic bags, foam materials, an identification accuracy of greater than 95% and a single frame processing time of less than 100 milliseconds.
  7. 7. The intelligent recovery environmental monitoring system of claim 5, wherein the AI processing unit is further capable of assessing a quality status of the recovery by analyzing the image characteristics, comprising: evaluating the degree of wear and the integrity of the recovered matters through surface texture analysis to generate an integrity score; detecting pollutant on the surface of the recycled material through characteristic recognition, including greasy dirt, label residue and foreign matter adhesion, and calculating the recycling rate of the recycled material according to the cleanliness score and the integrity score.
  8. 8. The intelligent recycling environmental monitoring system of claim 1, wherein the environmental assessment module comprises: the carbon footprint calculation unit is used for calculating the carbon emission reduction of the recovery behavior according to the identified recovery type, the estimated weight and the mass state and by combining a preset material carbon emission coefficient; the quality evaluation unit generates a quality score of the recovery according to the image characteristic analysis result; and the environmental protection integral calculation unit is used for integrating multidimensional indexes such as quantity, quality, classification accuracy, scarcity and the like to calculate environmental protection integral.
  9. 9. The intelligent recycling environmental monitoring system according to claim 8, wherein the carbon footprint calculation unit maintains a material carbon emission coefficient database, stores raw production carbon emission and recycling carbon emission of various materials, and the calculation formula of carbon emission reduction is: Carbon emission reduction = recycle weight x (raw production carbon emission coefficient-recycle carbon emission coefficient) ×renewable availability Wherein the weight of the recycle is obtained by image analysis in combination with density parameter estimation or by direct measurement by an integrated load cell.
  10. 10. The intelligent recycling environmental monitoring system of claim 8, wherein the carbon footprint calculation unit is further capable of calculating a resource conservation amount, comprising: savings Dan Youliang = PET plastic weight x petroleum consumption coefficient; water saving amount = paper weight x water consumption coefficient; Ore amount saved = metal weight x ore consumption coefficient; and the consumption coefficients are obtained by inquiring a preset database.
  11. 11. The intelligent recycling environmental monitoring system according to claim 8, wherein the environmental integral calculation unit calculates environmental integral by using a multidimensional evaluation model, and a calculation formula is: Basic integral = recycle weight x material integral coefficient x 10; Quality addition coefficient = cleanliness score x integrity score/10000; mass integral = base integral x (mass addition coefficient-1); Behavioral addition integral = classification correct score + high quality recovery score + rare material score; Total integral = base integral + mass integral + behavioral addition integral; wherein, the classification is correctly divided into 10-20 minutes, the high-quality recovery is divided into 10-30 minutes, and the scarcity material is divided into 20-100 minutes.
  12. 12. The intelligent recycling environmental protection monitoring system of claim 1, further comprising an environmental protection anomaly detection module, wherein the environmental protection anomaly detection module is configured to identify environmental protection anomaly scenes of a site in real time, and trigger hierarchical early warning according to anomaly types, the environmental protection anomaly scenes comprising: a pollution leakage scene, namely identifying ground pollutants, liquid leakage and water pollution; A smoke abnormal scene, namely identifying abnormal smoke, dust and fire; the ultra-high stacking scene is that the garbage stacking height exceeds a safety threshold value; And (3) the illegal operation scene is that the staff is identified to not wear protective equipment, and the illegal dumping and mixed throwing are carried out.
  13. 13. The intelligent recycling environmental protection monitoring system according to claim 12, wherein the environmental protection anomaly detection module adopts an anomaly detection algorithm based on deep learning, learns the feature distribution of a normal scene by training a self-encoder, calculates the difference between an input image and a reconstructed image as an anomaly score, and determines that the anomaly is abnormal when the anomaly score exceeds a preset threshold.
  14. 14. The intelligent recovery environmental protection monitoring system of claim 12, wherein the environmental protection anomaly detection module has an anomaly response time of less than 3 seconds and a false positive rate of less than 5%.
  15. 15. The intelligent recycling environmental monitoring system of claim 12, wherein the environmental anomaly detection module implements a hierarchical early warning mechanism comprising: Yellow early warning, namely slightly abnormal, wherein the abnormal score is between a threshold value T1 and a threshold value T2, and the action is recorded as a log and is parallel to a daily inspection plan; Orange early warning, namely moderate abnormality, abnormal score between a threshold value T2 and T3, action is to push a notification to a site manager, and rectification is required within 12 hours; Red early warning, namely, serious abnormality, abnormal scoring exceeding a threshold T3, and actions of immediately alarming, automatically recording video evidence, pushing multimedia notification, reporting a supervision platform and starting an emergency plan; Wherein T1 < T2 < T3, and the threshold is dynamically set according to the anomaly type and scene characteristics.
  16. 16. The intelligent recycling environmental protection monitoring system according to claim 15, wherein when the early warning is triggered, the system automatically records videos of 10 seconds before and after the early warning occurs as evidence, generates an early warning report containing a video screenshot, an abnormal position, an abnormal type and an abnormal time, and pushes the early warning report to a mobile terminal of a manager through a network.
  17. 17. The intelligent recycling environmental monitoring system of claim 1 wherein the data storage module comprises: the local storage unit is used for storing image data, identification results and environmental protection evaluation data by adopting high-speed SSD (solid state drive) storage; The block chain certification unit is used for storing the hash value of the key environment-friendly data in a uplink manner to form an environment-friendly record which cannot be tampered.
  18. 18. The intelligent recycling environmental monitoring system of claim 17, wherein the blockchain certification unit employs a federated chain architecture, and is commonly maintained by recycling enterprise nodes, regulatory agency nodes, third party audit agency nodes, and the uplink data comprises: The delivery link comprises a SHA-256 hash value, a recycle type, a number, a time stamp and a user ID of a delivered image, and a collection link comprises a hash value of a loaded image, GPS track data, transportation time and a vehicle ID; the sorting link comprises a sorting process image hash value, sorting accuracy, various category numbers and time stamps, and a regeneration link comprises a regeneration product image hash value, actual carbon emission reduction, resource utilization rate and time stamps; And each link data form chain storage, and the subsequent link data comprise hash references of the preceding link data, so that the non-tamper property and traceability of the data are ensured.
  19. 19. The intelligent recovery environmental monitoring system of claim 18, wherein the system generates a full-process environmental traceability report comprising: environmental protection characteristic data (types, quantity, quality, carbon emission reduction and the like of recovered matters) of each link; comparing and analyzing expected environmental benefit and actual environmental benefit, and proving block chain certification (transaction hash, block number and timestamp); The traceability report can be used for inquiry and audit by a supervision department and can be used as a basis for environmental protection subsidy reporting and carbon transaction authentication.
  20. 20. The intelligent recycling environmental monitoring system of claim 1, wherein the display interaction module comprises: The system comprises a user display screen, an integral feedback unit, a user display screen, a control unit and a control unit, wherein the user display screen adopts a high-definition touch screen to display recovery benefit information in real time; The recycling benefit information displayed by the user display screen comprises a recycling object list, carbon emission reduction, resource conservation amount, environmental protection integration and accumulated environmental protection contribution, and the recycling benefit information is displayed in a graphical mode, for example, the recycling benefit information is equivalent to planting X trees and saving X water.

Description

Intelligent recycling environment-friendly monitoring system and method based on computer vision Technical Field The invention relates to the technical field of computer vision technology and environmental protection monitoring, in particular to an intelligent recycling environmental protection monitoring system and method based on computer vision, which are particularly suitable for the situations of recycling resource recycling enterprises, garbage disposal sites, intelligent recycling equipment, environmental protection supervision departments and the like which need to monitor environmental protection indexes in the recycling process in real time. Background With the promotion of the national 'two carbon' strategy and the development of circular economy, the renewable resource recycling industry is coming to grow rapidly. Environmental monitoring becomes a core requirement of recycling enterprises, and is mainly used for proving environmental protection value (carbon emission reduction and resource saving) of recycling behaviors, meeting supervision requirements of environmental protection departments, acquiring government environmental protection subsidies and carbon transaction benefits, improving ESG (environmental, social and governmental) ratings of enterprises, displaying environmental protection contributions to the public and the like. At present, environmental protection monitoring of recycling enterprises mainly relies on manual recording and sampling detection, and has the following prominent problems: First, environmental data lacks objectivity and credibility. The environmental protection data is mainly manually filled, objective monitoring means are lacked, the authenticity of the data is difficult to verify, and the environmental protection benefit evaluation is inaccurate. For example, when recycling enterprises report carbon emission reduction to environmental protection departments, the provided data lacks third party evidence and has insufficient public confidence. Second, environmental protection monitoring coverage rate is low, has the supervision blind area. Traditional monitoring relies on manual inspection or fixed sensors, cannot cover the whole area, pollution events are not found timely, and environmental protection illegal behaviors are difficult to evidence. For example, certain recycling sites steal sewage at night, and due to lack of real-time monitoring, it is difficult to trace back and obtain evidence afterwards. Thirdly, the environmental protection value of the recycled material is difficult to evaluate. The recycled materials with different materials and different states have large differences in environmental protection values (carbon emission reduction and resource saving), and the environmental protection benefits of single recycling behaviors cannot be accurately evaluated. For example, a user has delivered a batch of waste paper, and the enterprise cannot tell the user how much carbon emissions the batch of waste paper specifically reduces, and the environmental incentive is poor. Fourth, environmental protection is difficult to trace to the source, and responsibility is difficult to define. The whole flow of the recycled materials from the process of throwing, transporting, sorting and regenerating lacks monitoring and recording, and the responsibility links are difficult to trace when the environmental protection problem occurs. For example, if a certain batch of recovered material is detected to be out of standard during the regeneration process, it cannot be determined which link is mixed. Fifth, environmental protection monitoring cost is high, and medium and small enterprises are heavy in burden. Professional environment-friendly monitoring equipment (an air quality monitor, a water quality analyzer and the like) is high in price, a set of complete environment-friendly monitoring system needs to be put into 50-100 ten thousand yuan, the medium and small recycling enterprises cannot bear the load, and environment-friendly monitoring flows in a form. The limitations of the prior art are mainly characterized in that (1) a traditional sensor monitors a dead zone, only monitors specific environmental protection indexes (such as PM2.5 and COD) and cannot comprehensively evaluate environmental protection conditions of a recycling site, cannot identify the classification quality of garbage in the site, cannot monitor whether the treatment process of recycled materials is standard and cannot evaluate the environmental protection value of single recycling behavior, (2) manual inspection efficiency is low, cost is high, inspection frequency is low (usually once in each month or each quarter), 7×24 hours of monitoring cannot be realized, labor cost is high, selective law enforcement is easy to occur, (3) intelligent environmental protection value evaluation is lacking, carbon emission reduction amount cannot be automatically calculated according to the types and quali