Design Of Image Edge Detection System Based On FPGA

With the rapid development of technology,video surveillance systems are now playing an increasingly important role in people's lives.Combining image processing algorithms can achieve target recognition and tracking functions,and edge detection algorithms are an important foundation for subsequent complex image processing algorithms.Therefore,the accuracy and real-time detection have a great influence on the final result.Due to the characteristics of FPGA pipeline and parallel computing,the edge detection algorithm can be well implemented while meeting real-time and high-speed requirements.Most edge detection algorithms use fixed thresholds,leading to unsatisfactory results,and constant testing is required to find a suitable threshold.This paper designs an adaptive threshold edge detection system.The main research contents are as follows:1.Check the relevant domestic and foreign literature to understand the development of edge detection technology,introduce some traditional detection algorithms,and improve the Sobel algorithm: first,a new boundary point filling scheme is used,and then the traditional horizontal and vertical direction The convolution template is expanded to 4 directions,the direction determination function is added,and then the adaptive threshold is calculated by histogram statistics,and finally the final result is obtained through the connection module.3.Design an improved edge detection algorithm.First,before the fast median filter,the boundary point filling module is designed,then the median filtered data is sent to the convolution module,and then the adaptive threshold module is designed and implemented,and finally sent to the connection module to get the detection result.At the same time,modelsim were used to simulate and verify each module.4.Build a hardware experiment platform,use Quartus software to compile and synthesize the code,burn the sof file into the FPGA,observe the comparison results,and find that the improved edge detection effect is better than before,and there is no need to manually set the threshold.