改进的HOG特征检测视频中的跑步动作
首发时间:2009-12-22
摘要:本文介绍了HOG特征结合二分类SVM方法检测视屏中的跑步动作,将传统的用于目标识别的HOG方法进行了改进,同时将其应用于作为检测跑步动作的主要特征,经过实验检测,确实能达到较好的效果。本文主要分为5个部分,分别进行了行为检测简介,改进的HOG方法介绍,SVM介绍,实际实验流程,最后给出了实验结果,并提出了改进意见。
For information in English, please click here
Improved HOG for running Action detection in the video
Abstract:This paper introduce a new method which employs the HOG and SVM,based the enhanced traditional Histogram Of Gradient which only be used to recgonize object.Mainly use the HOG features of Object, we can supervise the running action in the video.In fact,in our experiment, we find it is a good method. All the paper compose of 5 parts,including Action detection,HOG introduction(enhanced), SVM introduction,experiment details,and in the end ,we give some experiment results,and some suggestion for improvement.
Keywords: HOG SVM training classification
基金:
论文图表:
引用
No.3786250626312614****
同行评议
共计0人参与
勘误表
改进的HOG特征检测视频中的跑步动作
评论
全部评论0/1000