图像边缘特征提取是视频跟踪和目标提取的关键步骤,其性能和处理时间直接影响了后续图像处理的性能及视觉系统的整体响应时间。本文研究了一种基于人眼固视微动的实时图像边缘提取算法,针对算法具有边缘识别能力强,但运算量大的特点。通过采用并行处理结构及流水线技术在FPGA内实现了该算法,试验结果说明该算法在FPGA内实现具有与视频流同步的实时性,并且边缘提取的效果很好。在图像目标实时跟踪系统中应用,极大提高了目标的跟踪精度和稳定性。

Image edge extraction is the key step of video tracking and target identification.The performance and processing time directly influence the follow-up image processing performance and the overall visual system response time.A real-time infrared image edge extraction algorithm of eye movement mechanism is presented.The algorithm has features of strong edge recognition,but the computation is complex.Through the use of parallel processing structure and the pipeline technology,the algorithm is realized in a Field Programmable Gate Array(FPGA).The results show that the algorithm implements synchronization of real-time video streaming in a FPGA and have a good image edge extraction.Application of the image in real-time tracking system greatly increases the target tracking accuracy and stability.