AN ALGORITHM FOR MULTIPLE OBJECTS TRACKING AND BEHAVIORS RECOGNITION AND DESCRPTION

摘要：

In this paper, a new behavior recognition algorithm based on multiple objects tracking and matching of foreground blobs and the predict positions and sizes of the objects are present.The algorithm uses correlation matrix and two matching chained list (object-blob matching chained list and blob-object matching chained list) to corresponding blobs and objects, and match blobs and objects in various conditions.Labeled area is used to help to recognize behaviors which are defined with the help of the tracking results of the objects.Finally, behaviors are described in the sentences of natural languages.

关键词： Multiple Objects Tracking Correlation Matriz Matching Chained List Behavior Recognition

作者: Kejia Bai

作者单位: School of Computer Science,Guangdong Polytechnic Normal University Guangzhou 510655,P.R.China

会议类型: 国际会议

会议名称: 2009 International Symposium on Computer Science and Technology(2009 中国宁波国际计算机科学与技术学术大会)

会议地点: 宁波

会议语种:英文

页码: 262-266

在线出版日期: 2009-11-20（万方平台首次上网日期，不代表论文的发表时间）