Recognition of Natural Road Sign Based on the Improved Curvature Feature

• 作者: Hao Zheng Weiwei Chen Yanqing Wang
• 来源期刊: 国际计算机前沿大会会议论文集
• 学科: 社会科学
• 关键词: FEATURE MATCHING FEATURE extraction ROAD SIGN CURVATURE algorithm
• 年，卷（期）: 2017,（1）
• 页码范围: 170-171
• 页数: 2页
• 分类号: C5

摘要

To solve the recognition of road sign with an intelligent vehicle in vision-based navigation,road sign extraction and matching techniques required in outdoor scene was proposed in this paper.The method of the improved curvature based on feature extraction and binary description took the advantage of reasonable features distribution to overcome the problems of traditional features uneven distribution.Binary description method was represented to solve the real-time problem of feature matching.Through the validity and real-time performance of different algorithms are compared by experiments and indicate that the method can not only overcome negative influences from the disturb of non-targets,while spending on average only 46 ms processing each frame,but also meet the requirements of robustness,real-time,and accuracy.