摘要
提出了一种新的基于FPGA的立体图像差异性算法,它以块匹配算法为基础,根据FPGA的特点,对图像相关性的公式进行设计优化,并结合穷尽方式搜索和预测方式搜索,提高算法的执行速度.设计基于FPGA的立体图像差异性算法IP核,充分利用FPGA独特的并行处理机制和强大的运算能力以提高系统的处理速度和性能.系统测试结果表明,基于FPGA的立体图像差异性算法,可以达到每秒33帧的处理能力,处理速度能够达到PC机的二百倍以上,具有较好的实时性;且能够连续处理500帧图像数据,具有较好的稳定性.
A new stereo image disparity algorithm was proposed based on FPGA.Based on the block matching algorithm and according to the characteristics of FPGA,the formula of image correlation was optimized.Moreover,by combining exhaustive search and prediction search method,the execution speed of the algorithm was improved.The Stereo image disparity algorithm IP core was designed,and the processing speed and performance of the system could be improved by making full use of the FPGA’s unique parallel processing mechanism and powerful computing ability.The test results show that the stereo image disparity algorithm based on FPGA can reach the processing capacity of 33 frames per second and the processing speed can reach more than 200 times of PC,with good real-time performance.It can process 500 frames of image data continuously and has good stability.
作者
李贞妮
王骄
李晶皎
王泽坤
LI Zhen-ni;WANG Jiao;LI Jing-jiao;WANG Ze-kun(School of Information Science&Engineering,Northeastern University,Shenyang 110819,China)
出处
《东北大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2020年第6期778-783,共6页
Journal of Northeastern University(Natural Science)
基金
国家自然科学基金重点项目(61836011)
中央高校基本科研业务费专项资金资助项目(2020GFYD011,2020GFZD008)。