Research And Application Of Computer Vision In Violation Identification System

With the enhancement of national strength and the prosperity of the people,the number of private cars and private car travel are gradually replacing the way of public transportation.Traffic violations are the core concern of modern transportation,and also the key to ensure the safety of people’s property and personal safety.In reality,traffic law enforcers are often faced with irresistible factors such as weather,unable to carry out the road face-to-face supervision.They need to respond to the traffic violations quickly and timely.They need the violation identification system to assist the traffic law enforcers in identifying and recording the traffic violations.At present,some violation photo systems on the market are too single and accurate the degree needs to be improved,and it will be greatly influenced by some external factors when dealing with interference events,so it is more practical to use computer vision in the problem of violation identification.First of all,this paper describes the research trends of computer vision at home and abroad.With the current computing power and network transmission capacity of computer,the problem of computer vision and machine learning can be solved.Further research on the algorithm of image preprocessing,including the basic representation of image,morphological operation,edge detection and line detection Among them,the most important are morphological operation,edge detection and line detection,which play a key role in the subsequent illegal behavior recognition.Then it analyzes the common violations in daily life,such as illegal parking,illegal lane changing,illegal retrograde and the wearing of safety belt,analyzes and selects the design monitoring scheme in turn,makes a variety of comparisons for the vehicle recognition algorithm,and finally selects the most compromised Yolo algorithm on the recognition effect and speed to identify the vehicle,and then designs the algorithm for identifying violations,but only However,it is not enough to recognize vehicle violations.It is necessary to record the vehicles violating the regulations,which involves vehicle license plate recognition,including license plate location,license plate character segmentation and single character recognition.This paper improves the detection method of illegal parking based on duty cycle and applies it to the system of illegal parking recognition.Using the method of vehicle identification and duty cycle to identify the vehicles that stop illegally,which improves the effectiveness of illegal detection.Finally,a large number of vehicle data sets are collected and labeled,and then the recognition weights are trained.Finally,the recognition algorithm is compared to verify the effectiveness of the recognition scheme.At the same time,a violation recognition system is developed according to the actual situation,and the research algorithm is effectively applied in the system.