Research On Detection Method Of River Floating Foreign Objects Based On Machine Vision

The hydrological examination provides the basis for water conservancy construction,flood control and drought resistance,and water resource management by observing hydrological elements.The flow measurement of the lead fish is an important item in the hydrological examination.During the flow measurement,river floating foreign objects colliding with the lead fish and its suspended flow meter will cause the flow meter to be damaged.In serious cases,the entanglement of foreign objects causes the cable channel carrying the lead fish to collapse.River floating foreign objects seriously interfere with the normal work of the hydrological examination,causing damage to the examinationing facilities and even endangering the safety of the staff.Therefore,the river floating foreign objects are detected and the threat level of the foreign objects is judged,and early warning and obstacle avoidance are performed during the flow measurement of the lead fish cable channel to ensure the flow measurement safe and reliable is of great significance.Compared with the manual observation method,the river floating foreign objects detection method based on machine vision technology has greater application value for realizing remote automatic flow measurement,which meets the needs of the national”smart hydrology” construction.Moving target detection is one of the hotspots of current machine vision research.This article mainly studies the river floating foreign objects detection method and the threat level model of floating foreign objects to lead fish.Firstly,the characteristics and hazards of river floating foreign objects are introduced,the significance of research on foreign objects detection in rivers is explained,the feasibility and advantages of moving object detection methods in foreign objects detection in rivers are analyzed,and the research status of foreign objects detection at home and abroad is summarized.Secondly,image preprocessing algorithms such as image graying,image filtering,and image enhancement are studied to enhance the clarity of the river scene image and highlight the target features,which provides the basis for the subsequent detection of river floating foreign objects.Then I introduce the moving target detection algorithm,and analyze the characteristics of each algorithm.Improves on the problem of poor detection effect in the background scene of the river based on the background template-based visual background extraction algorithm(Vi Be).Proposes an adaptive threshold the method and the method of parameter negative feedback to improve the adaptability of the background model to the dynamic background,and improve the robustness and detection accuracy of the algorithm by fusing space-time features.The improved method is used to reliably detect the presence of foreign objects in the process of lead fish flow measurement in real-time video scenes of river courses,and to locate and geometrically analyze the foreign objects.Then I use the You Only Look Once(YOLO)algorithm to recognize the types of river floating foreign objects,and perform a lightweight operation on the YOLO algorithm to improve the recognition speed.Finally,according to the basic characteristics of foreign objects and the working status of the lead fish,a foreign object threat level model is set to determine the threat degree of the foreign objects to the lead fish and issue a warning of the foreign objects impacting the lead fish to perform the obstacle avoidance operation of the lead fish.The experiments on the CDNET standard dataset and the river floating foreign objects dataset show that the improved method in this paper achieves a good balance between real-time performance and accuracy based on the effective detection of foreign objects and the acquisition of basic characteristics of foreign objects,and solves the detection of moving targets.The algorithm detects the problem of low accuracy in river scenes with a dynamic background.By identifying the type of foreign objects,it can better determine the threat level of the foreign objects to the lead fish,and avoid the damage of the flow measurement facility to the foreign objects of the river,and ensure the safety of the flow measurement work.