QQ客服 官方微博 反馈留言

	Adaboost video tracking
	Jia Jingping ; Chai Yanmei ; Zhang Feizhou
	2008
关键词	SELECTION
英文摘要	A new approach of tracking objects in image sequences is proposed, in which tracking is seen as a binary classification problem. For each incoming image frame, a likelihood image for the object is created according to the classification results of pixels by a Adaboost feature classifier. In the likelihood image the object's region turns into a blob. The scale of this blob can be determined by the local maxima of differential scale-space filter. We employ the QP_TR trust region algorithm to search for the local maxima of the multi-scale normalized Laplacian filter of the likelihood image so as to locate the object as well as determine its scale. The object's appearance change is dealt with in the update step of the feature classifier. Based on the tracking results of sequence examples, the novel method has been proven to be capable of describing the object more accurately and thus achieves much better tracking precision.; Computer Science, Artificial Intelligence; Computer Science, Theory & Methods; Engineering, Biomedical; Engineering, Electrical & Electronic; Imaging Science & Photographic Technology; CPCI-S(ISTP); 0
语种	英语
内容类型	其他
源URL	[http://ir.pku.edu.cn/handle/20.500.11897/312067]
专题	地球与空间科学学院
推荐引用方式 GB/T 7714	Jia Jingping,Chai Yanmei,Zhang Feizhou. Adaboost video tracking. 2008-01-01.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。