摘要
针对篮球视频中运动员的检测识别问题,提出一种基于AdaBoost分类器的检测方法 .首先,从视频中获取有用帧,并通过SampleCreator软件来标记运动员,提取出全身和上半身矩形图像.然后,基于积分图技术从对象图像中提取Haar特征.接着,利用AdaBoost算法选择出具有较强分类性能的特征,训练一系列的弱分类器,并将其进行级联来构建最终的强分类器.最后,通过强分类器对Haar特征进行判别,从而检测图像中的运动员.实验结果表明,该方法能够准确检测并识别视频中的运动员.
For the issue that the athlete detection and recognition in basketball video,a detection method based on AdaBoost classifier is proposed.Firstly,the useful frames are obtained from the video,and through the SampleCreator software to mark the athletes,so as to extract the rectangular image of whole body and upper body.Then,the Haar feature is extracted from the object image based on the integral graph technique.After that,the AdaBoost algorithm is used to train a series of weak classifiers which are used to construct the final classifier.Finally,the Haar features are distinguished by the strong classifier to detect the players in the image.Experimental results show that the proposed method can accurately detect and identify the players in the video.
出处
《湘潭大学自然科学学报》
CAS
北大核心
2016年第4期85-89,共5页
Natural Science Journal of Xiangtan University
基金
湖北省高校教学研究项目(2013160)
