The Research Of Illegal Parking Detection And License Plate Recognition System

With the advancement of technology and the improvement of people’s livingstandards, the number of motor vehicles is also growing rapidly, thus generating a lotof traffic violations, traffic management systems therefore need to constantlyintroduce new technology to improve traffic management in real-time, reliable,efficiency, to provide effective protection for the community and orderly traffic.Therefore the Intelligent transport system appeared, which can play a role in acomprehensive and large-scale, real-time, high accuracy and efficient, integratedtransport and management system.First of all, the thesis studies several video image preprocessing methods, suchas image filtering, edge detection, morphology, thresholding segmentation, etc. Thesemethods lay the foundation for follow-up work of the subsequent image processingand analysis.Due to the phenomenon of illegal parking is becoming more serious every day,This thesis is designed based video surveillance to detect illegal parking and licenseplate recognition system illegal parking identification system, which is composed ofthe following components: extracting foreground objects from a video sequence,interference filter out the noise and non-motorized vehicles, moving vehicle tracking,stationary vehicle detection, illegal parking detection and judgment, violation of thevehicle license plate recognition.In the foreground object extraction, using a Gaussian mixture backgroundmodeling method based on background subtraction method, which overcomes robustlight, twigs and other effects caused by the shaking. Interference noise andnon-motorized vehicles using the object-level region segmentation method using theduty cycle to filter out objects within the region. Detection of moving vehicle usingthe Kalman filter is introduced in the Camshift algorithm to achieve better tracking ofthe vehicle, which can effectively solve the situation of the background color issimilar to the target or the target is blocked seriously. The stationary vehicle isdetected using the average displacement of the vehicle in a certain time. The detectionof illegal parking of vehicles is judged by the distance of vehicle centroid with noparking area centroid and residence time.License plate recognition was detected on the basis of illegal parking. This thesissplit from the license plate location, character and license plate character recognitionalgorithms to analyze three steps and make some improvements, for example, thethesis made some improvements to the defects of license plate tilt correction and thecharacters are not communicating, and achieved better results, ultimately identify thelicense plate, thus forming a complete illegal parking detection and automatic licenseplate recognition system. Research of this system can promote the development ofintelligent traffic management system in a certain significance.