The Implementation Of Sobel Edge Detection Algorithm Based On HLS And FPGA

As an important form of carrying information,image is playing a significant role in our life.Edge is a basic feature of an image;the purpose of edge detection is to reduce the amount of data in an image while preserving the shape information of objects.Edge detection is a critical step for further image processing,and it has a wide range of applications in many fields.The feature of Central Processing Unit that its instructions are executed serially leads to it cannot meet the requirements of large data,high speed,high stability,it is an inevitable choice to design a special circuit for hardware acceleration.Field programmable gate arrays(FPGA)are best suited to implement image processing algorithms with the reason that it provides infinite reprogram ability and high parallelism.The image processing algorithm involves a lot of division,square root index,so it is very difficult to realize using HDL.In addition,designing in RTL has the problems of system maintainability.The high level synthesize technology can solve those problem.It is a new choice of embedded system design based on FPGA.This paper starts with the research on different classical edge detection algorithms,and makes Sobel algorithm compare with others.The hardware and software architecture of video processing based on ZYNQ is discussed then.After discussing the development process of embedded system,it designs a Sobel filter IP core based on HLS method and makes some optimizations in further.In order to verify the effectiveness and efficiency of the design,a real-time Sobel edge extraction system is designed.The system uses USB camera for real-time video capture,and then the Sobel edge extraction processing results are displayed in real time through the HDMI device.The experimental results show that the efficiency of Sobel edge extraction using FPGA is 20-100 times than using ARM processor.It greatly improves the speed of the algorithm;The hardware acceleration effect is remarkable and the greater the amount of data,it has greater effect.