盲道识别系统算法设计及FPGA实现

The Design of Blind Sidewalk Detection Algorithm Based on FPGA

将FPGA并行处理技术与数字图像处理技术相结合,采用自顶向下分块设计原则提出一种适于FPGA实现的盲道识别算法.整个算法设计根据盲道与人行道有无色差分成2大模块并行处理,其中对有色差部分提出了基于H色相值的盲道识别算法,对无色差的部分提出了形态学优化的边缘检测算法.本设计提出的基于形态学优化的边缘检测包括sobel算子边缘检测,优化的形态学膨胀处理及hough直线检测出盲道边界线三部分.仿真结果表明本系统对盲道线识别具有良好的实时性与精确的识别精度.

This paper discusses a blind sidewalk detection algorithm based on FPGA parallel processing tech- nology and digital image processing technology. Depending on the colors of blind sidewalk and pavement, the algo- rithm includes two major modules which are processed in parallel. In one module, blind sidewalk is detected by the value of the hue when the colors of blind sidewalk and pavement are significantly different. In another module, blind sidewalk is detected by edge detection algorithms optimized by mathematical morphology, when the colors be- tween blind and sidewalks are similar. The proposed design optimization based on morphological edge detection in-cludes three parts： sobel operator edge detection, morphological dilation image processing and Hough transform for line detection. The simulation results show that the system for blind sidewalk detection has a good real -time and accurate recognition accuracy.