摘要
提出了一种基于最优次模式(OSPA)距离的路标识别算法.算法引入Markov随机场建立噪声的待检图像,然后采用条件迭代算法(ICM)恢复图像,进一步提取路标边缘点.这些边缘点作为特征点;该特征点看作为待检特征点集,计算其和标准路标库图像之间的OSPA的距离,以此来识别待检路标.分析表明,该算法对于路标的形态识别具有明显的优势,最后分析了图像尺寸大小、特征点数量对OSPA距离的影响.
In this paper, a recognition algorithm for road signs based on the optimal sub -pattern assignment (OSPA) metric is proposed. The noisy road sign picture is gained by using Markov random fields. The true picture is then recovered by using iterative condition method (ICM). Based on the recovered picture, the marginal points are derived and seen as pending characteristic points. Then, the OSPA metric is proposed to evaluate the distance between standard characteristic points and pending characteristic points. The standard one which has the smallest OSPA distance is judged to be the correct class. The final results show that proposed algorithm gives a'better recognition. Finally, the influence of picture size, the number of characteristic points on OSPA is analyzed.
出处
《哈尔滨师范大学自然科学学报》
CAS
2017年第2期55-57,共3页
Natural Science Journal of Harbin Normal University
基金
浙江省自然科学基金资助项目(LY15F030020)
杭州电子科技大学2016年高等教育研究资助项目资助."本科生实验室科研创新研究与实践"(YB201662)