文献类型：期刊文献

中文题名：基于FPGA嵌入式视觉的导盲算法实现

英文题名：Blind-guiding Algorithms Based on FPGA Embedded Vision

作者：孙鑫宇[1];储辉[1];杨天歌[1];高天维[1];王嵘[1];万永菁[1]

机构：[1]华东理工大学自动化系,上海200237

年份：2020

卷号：39

期号：1

起止页码：79

中文期刊名：计算技术与自动化

外文期刊名：Computing Technology and Automation

收录：CSTPCD

基金：国家自然基金项目资助项目(61872143);教育部产学合作协同育人资助项目(201801315013)。

语种：中文

中文关键词：FPGA;盲道识别;嵌入式视觉

外文关键词：FPGA;blind sidewalk identification;embedded vision

摘要：项目在FPGA平台上实现了一系列的图像处理算法。通过HSV色彩空间上的阈值判断、形态学滤波等方法识别出了一张图像中的盲道成分,加以一系列的分析判断,得出盲道的详细情况,并充分发挥了FPGA的高速并行特点,实现了每秒千帧的超高速处理。结果表明,系统具有很高的识别准确度与很强的环境适应性。
Our project implements a series of image processing algorithms on the FPGA platform.Through the morphological filtering on the HSV color space and threshold judgment,the blind sidewalk components in an image are identified,and a series of analysis and judgment are made to obtain the details of the blind sidewalk,and the high-speed parallel characteristics of the FPGA are fully utilized to realize ultra-high speed processing of thousands of frames per second.The final result shows that the system has high recognition accuracy and strong environmental adaptability.