激光与光电子学进展, 2015, 52 (5): 051004, 网络出版: 2015-05-07
基于区域特征匹配的扩展目标高精度跟踪 下载: 586次
High-Precision Extended Object Tracking Based on Region Feature Matching
摘要
为了能够更精确的跟踪扩展目标,提出了一种基于最大稳定极值区域(MSER)特征匹配的高精度扩展目标跟踪方法,该方法通过对扩展目标图像中的一个区域进行抽样式多阈值二值化处理提取出MSER,再通过对MSER 面积变化率条件的判断保证质心提取精度,并与上一帧MSER 进行相似度匹配找到下一帧的MSER,将其质心作为跟踪坐标。实验结果表明该方法对于存在MSER 特征的扩展目标具有更精确和更稳定的跟踪能力,在面积变化率一定的条件下,其跟踪精度不低于0.1 pixel,且计算时间也减少了约3/4。
Abstract
In order to track extended object more precisely, a high-precision extended object tracking method is proposed based on maximally stabel extremal regions (MSER) feature, according to binarization with sampling multithreshold, the MSERs are detected in a region of extended object, if area change rate of MSER meets centroid precision condition, the MSER is picked out. And MSER in next frame are extracted and matched with the MSER in previous frame, the centroid of the MSER matched is next tracking position. Experimental results show that the proposed tracking method can improve the stabilization and precision of extended object tracking, if area change rate of MSER meets the certain condition, the tracking precision is better than 0.1 pixel, and the time consumption reduces by 3/4.
张建伟. 基于区域特征匹配的扩展目标高精度跟踪[J]. 激光与光电子学进展, 2015, 52(5): 051004. Zhang Jianwei. High-Precision Extended Object Tracking Based on Region Feature Matching[J]. Laser & Optoelectronics Progress, 2015, 52(5): 051004.