摘要
在序列图像中进行目标跟踪是计算机视觉、图像处理和模式识别领域里非常活跃的课题 .采用 Hausdorff距离模板匹配的方法具有计算量小、适应性强的特点 .为了能对复杂背景 (包括运动背景或移动镜头 )情况下的序列图像进行准确的跟踪 ,综合考虑了图像边缘的位置信息和方向信息 ,对模板匹配和模板更新的策略作了改进 ,与原有方法相比 ,目标跟踪的准确程度和算法的效率有了显著提高 .
This paper described techniques to track nonrigid object moving in a complex scene. A new version that incorporates both location and orientation information was used in the model-match and model-update process in the techniques. A high efficient searching strategy of locating object in an image was also proposed. The results indicate that these techniques can perform well in the conditions such as moving background and moving camera.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2001年第2期223-227,共5页
Journal of Shanghai Jiaotong University
