Design And Implementation Of Underlying IP Core For Smart Camera Image Processing Based On HLS

The smart camera is a highly integrated micro-machine vision system that is widely used in modern industrial production.Image preprocessing and local feature extraction are used as low-level image processing in smart cameras.The processing effect directly affects the reliability of subsequent high-level image processing.With the rapid development of science and technology,the resolution of image acquisition is increasing,and people's requirements for real-time image processing are getting higher and higher.Using FPGA to design the underlying IP core is one of the effective ways to improve image processing speed.Based on the Zynq-7000 development platform,this paper uses the high-level synthesis tool Vivado HLS(High Level Synthesis)to design the underlying IP core of smart camera image processing,which realizes the hardware acceleration of low-level image processing in smart cameras.This paper mainly studies the following aspects:(1)Analyze the principle of image filtering algorithm and local threshold binarization algorithm,find the commonality of these two types of image preprocessing algorithms,and complete the preprocessed IP core design by acquiring window data and processing image boundary pixels.(2)On the basis of image filtering using HLS,after analysis and design,the mean filtering,Gaussian filtering and median filtering are integrated into one IP core,which achieves the purpose of switching different filtering methods according to actual needs.(3)By studying the local features of the image,find the corner features and edge features suitable for HLS implementation.Firstly,the principle of common corner detection algorithm is analyzed.Based on HLS,the IP core design of Harris corner detection algorithm is realized.Then the principle of edge detection algorithm is analyzed,and the IP core design of Prewitt edge detection algorithm is realized.Finally,the image algorithm function verification is performed on the ZC702 development board,and the performance analysis is performed according to the synthesise report,including the software and hardware processing time of the image preprocessing algorithm and the local feature extraction algorithm,and the hardware resource utilization rate.The experimental results show that the image processing algorithm based on HLS is correct and meets the requirements of real-time processing.In addition,the local threshold binarized IP core is applied to the two-dimensional code recognition of smart cameras,which realizes the hardware acceleration of image preprocessing algorithm in two-dimensional code recognition.