摘要
针对现代工业需求,设计了一种基于ZYNQ的图像采集、畸变矫正、代价计算、代价聚合的双目视觉系统。实验结果表明,该系统仅以1.649 W功耗完成图像的采集矫正并最终得出视差图,计算速度较传统PC有极大提升,计算结果误差仅为0.121个像素。
According to the needs of modern industry,a binocular vision system based on ZYNQ is designed,which includes image acquisition,distortion correction,cost calculation and cost aggregation.The test results show that the system only completes the image acquisition and correction with 1.649 W power consumption and finally obtains the parallax map.The calculation speed is greatly improved compared with the traditional PC,and the calculation result error is only 0.121 pixels.
作者
张宏宇
焦新泉
贾兴中
Zhang Hongyu;Jiao Xinquan;Jia Xingzhong(State Key Laboratory of Electronic Technology Testing,North University of China,Taiyuan 030051,China;Key Laboratory of Instrument Science and Dynamic Testing,Ministry of Education,North University of China)
出处
《单片机与嵌入式系统应用》
2023年第6期52-56,共5页
Microcontrollers & Embedded Systems