| Summary: | 碩士 === 逢甲大學 === 資訊工程學系 === 105 === In recent years, the high population density and convenience of vehicles cause the demand of the parking lot or roadside parking increased. When the demand is not available, people usually park the car at will. This sometimes causes the troubles and accidents to other people because of shrinking the roadway or blocking the view of corners. For another case, someone may drive to sparsely populated areas and throw a lot of garbage, which will cause the environment pollution. So, the official organization have to configure temporary cameras to make them be punished .
In the thesis, we design a real-time and low-cost License Plate Recognition system on embedded platform with general camera, which can be deployed at roadside for temporary monitoring. The record of the recognized license plate number and the corresponding pictures and videos can be sent back to the server for further process. The proposed methods will first set the appropriate environmental parameters. Then, the character contour’s feature is used to find the license plate in the close distance and vertical edges are used to find it in the median distance. Next, perspective transform can convert the plate to face angle, and then unnecessary noises can be filtered. And characters can be separated according to its binary and contour value constraints. Finally, the old and new license plate characters can be identified by the prepared training data with KNN classifier.
In the experimental results, the overall recognition accuracy of the license plate number is 86.83%. The accuracy of finding out the license plate is 98.33%. The accuracy of separating characters is 96.69%. The accuracy of character or number recognition is 91.33%.
|
|---|
