摘要
现场可编程门阵列(Field Programmable Gate Array,FPGA)的并行处理能力能够极大提升数字图像处理效率。通过FPGA编程,在硬件层面提出一种实时图像边缘提取系统。对摄像头采集的视频图像实施灰度转化、高斯滤波、二值化等预处理操作,再利用Sobel算子进行边缘检测,提取图像边缘信息,最终经过VGA显示器实时输出。创新性地提出通过线性移位寄存器级联的方式实现矩阵边界填充,解决了部分边界值的不可用性;利用双RAM乒乓操作输出图像数据,解决了模块间速率不匹配问题;设计人为可调节阈值,过滤不同光照强度带来的噪声影响,边缘相关系数由0.63提高至0.91。系统实时性好,资源占用率仅为可用资源的0.67%~23.70%,在与大型FPGA图像处理工程对接中具有应用潜力。
The parallel processing capability of Field Programmable Gate Array(FPGA)can greatly improve the efficiency of digital image processing.Through FPGA programming,a real-time image edge extraction system was proposed on the hardware level.The video images collected by the camera were preprocessed by gray scale conversion,Gaussian filtering and binarization,and then the edge detection was carried out by Sobel operator to extract the image edge information,and finally the image edge information was output in real-time by VGA display.Innovatively proposed consist of a cascading linear shift register to realize matrix boundary filling to solve the unavailability of some boundary values;a double RAM ping-pong operation to output image data,to solve the problem of rate mismatch between modules;an artificially adjustable threshold to filter the noise caused by different light intensities,so that the edge correlation coefficient is increased from 0.63 to 0.91.The system has good real-time performance,and the resource occupancy rate is only 0.67%~23.70%of the available resources.It has application potential in docking with large-scale FPGA image processing projects.
作者
高宏玲
白昊明
余山
翟腾
郭青帅
GAO Hongling;BAI Haoming;YU Shan;ZHAI Teng;GUO Qingshuai(China Software Testing Center(Software and Integrated Circuit Promotion Center of Ministry of Industry and Information Technology),Beijing 100048,China;Telecommunication Science and Technology Research Institute,Beijing 100191,China)
出处
《工业技术创新》
2022年第3期112-119,共8页
Industrial Technology Innovation
