Research On Sobel Edge Detection Algorithm And FPGA Implementation

The rapid development of modern society cannot do without information processing,and the image is the largest part of the information obtained by human beings.Modern industry uses computers to process natural images,on the one hand,to facilitate human eyes to observe and recognize complex images,more is to achieve the purpose of computer automatic recognition and understanding of images.The edge of an image is the most basic feature of the image,which contains abundant internal information of the image.Therefore,edge detection is one of the key research tasks in image processing.However,with the pursuit of higher and higher image quality,the amount of image data that the computer needs to process is increasing day by day,and the traditional software processing methods can not meet the real-time requirements of many scenes.How to transmit and process high resolution video at high speed has become a new field of digital image research.The development of semiconductor technology brings a kind of semiconductor chip,field Programmable logic Gate array(FPGA),which can process image data in parallel at high speed.This paper designs an improved Sobel edge detection system based on FPGA,which mainly includes video acquisition,data cache,image processing and VGA display.OV5640 camera provides HD video data source.The frame cache of video data is completed by SDRAM memory chip,and the video data is output to VGA display after the completion of in-chip processing by FPGA.In this paper,aiming at the problems of inaccurate positioning and rough edges caused by fixed threshold in the classical Sobel edge detection algorithm,an edge detection algorithm combining eight-direction Sobel operator and eight-direction nonmaximum suppression is proposed.Firstly,the eight-direction convolution template is used to improve the accuracy of edge positioning,and then the eight-direction nonmaximum suppression method is used to refine the gradient image edge after the extended direction.Finally,the double-threshold edge discrimination method is used to obtain a more ideal image edge.Firstly,the result graph of the improved algorithm in this paper is obtained by Matlab software,and compared with the detection result graph of the traditional Sobel algorithm,the improved algorithm in this paper gets a better edge.The design of the system module adopts Verilog HDL language and the development of each module is completed on the Quartus Prime 17.1 software development platform.After the system design is completed,the image processing module is simulated by using the Questasim 10.6c simulation platform to verify the correctness of the timing logic.Finally,the sof file is downloaded to the FPGA and the detection results are observed on the monitor.Experiments show that the edge detection system in this paper can carry out real-time edge detection and achieve better edge results.