Research On Key Feature Recognition System Based On Agricultural Vehicles

In order to improve the accuracy of license plate recognition of agricultural vehicles and the management efficiency of public security organs,this paper analyzes the characteristic structure of agricultural vehicle license plate,and puts forward an effective method of license plate recognition of agricultural vehicles:(1)improve Canny edge detection algorithm.Compared with the traditional edge detection algorithm,Canny algorithm has better effect in noise reduction and edge protection,but in detail There are still some deficiencies in the recognition.In this paper,an improved Canny operator edge detection algorithm is proposed.The adaptive median filter with edge preserving denoising characteristics is used instead of Gauss filter.Experimental results show that this algorithm can recognize more effective edges than traditional edge detection;(2)Text detection algorithm based on stroke width transform fusion.In this paper,we propose a significant text feature detection method,which combines spectral residual significance detection and stroke width transformation detection,and apply this method to the MATLAB architecture system.The experimental results show that the performance of this algorithm is improved,and the algorithm speed and feature detection performance meet the expected requirements.using Matlab r2018 a to realize the license plate recognition system.The main steps of the system are: preprocessing the image of agricultural license plate,locating the agricultural license plate,character segmentation of license plate number,character recognition of license plate number.In the preprocessing,in order to accurately identify the license plate number,the optimized Canny algorithm is used in the edge detection,and a large number of agricultural license plates are tested.Around the license plate recognition technology,we also study(1)license plate recognition based on BP neural network;(2)template matching character recognition algorithm.The experimental results show that the recognition rate of the agricultural license plate recognition system studied in this paper is high,and the accuracy rate is 98.12%.