基于PCA-LBP算法的竞技赛技术动作行为识别模型  

Behavior Recognition Model of Competitive Action Based on PCA-LBP Algorithm

作　　者：蒋锐[1] 

机构地区：[1]九江学院体育学院,江西九江332000

出　　处：《科技通报》2016年第6期162-165,共4页Bulletin of Science and Technology

摘　　要：针对标准LBP算法在竞技赛技术动作行为识别的应用中还存在识别精度不高等问题。本文提出了一种基于概率和PCA降维优化LBP算法的竞技赛技术动作行为识别模型,首先根据应用概率论选定一种相似度量方式作为距离,然后按对应权重在相似度级别上融合,再以这种度量方式分类识别,最后采用PCA算法对LBP法提取的运动员特征进行降维处理,以降低算法的运算量。实验仿真结果表明,本文提出的改进LBP算法在竞技赛技术动作行为识别的应用中具有更高的识别精度。According to the low identified accuracy of the standard LBP algorithm in the event action recognition, a behavior recognition model of event technology is proposed based on probability and LBP with the optimization of PCA dimensionality reduction. First of all, a similar measure is selected as the distance according to the applied probability theory. Then according to the corresponding weight on the similarity level fusion, this measurement is for classifying and recognizing. Finally the PCA algorithm is adopted to athletes LBP method to extract feature dimension reduction processing, in order to reduce the computational complexity of the algorithm. Simulations show that the proposed improved LBP algorithm has higher identification accuracy in the event action recognition.

关 键 词：LBP算法 竞技动作 行为识别 PCA降维 概率论 运动员识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]