基于改进LSDA的人体行为识别算法
|
摘要
提出了一种基于改进的局部敏感判别分析(LSDA)的人体行为识别算法。首先,利用样本间的距离信息重置类内最近邻点集和类间最近邻点集的权值矩阵,保证在最大化目标函数时尽可能保持局部线性重构特性,解决小样本难题;然后,通过Gran-Schmidt正交化方法将特征矩阵进行正交化,可有效地保留原始空间中的高维空间,解决图像特征重建难题;最后在Weizmann行为数据库进行实验,平均正确识别率达到98.21%。试验结果表明,该算法具有良好的分类性能,在不同训练样本个数下,该算法均能获得较高的识别率。
A method of behavior recognition algorithm based on modified locality sensitive discriminant analysis is proposed. Firstly,the parameters of weight matrix of recently adjacent vertex sets in class or out class is resetted by the distance of samples to solve sample size problem. Secondly,the Gram- Schmidt orthogonalization is introduced to preserve higher dimensional space in the original space effectly. Finally,the average recognition rate of the proposed method is 98. 21% based on the Weizmann behavior database. The experimental results show that the proposed method is effective and feasible for behavior recognition.
A method of behavior recognition algorithm based on modified locality sensitive discriminant analysis is proposed. Firstly,the parameters of weight matrix of recently adjacent vertex sets in class or out class is resetted by the distance of samples to solve sample size problem. Secondly,the Gram- Schmidt orthogonalization is introduced to preserve higher dimensional space in the original space effectly. Finally,the average recognition rate of the proposed method is 98. 21% based on the Weizmann behavior database. The experimental results show that the proposed method is effective and feasible for behavior recognition.
引文
[1]雷庆,陈锻生,李绍滋.复杂场景下的人体行为识别研究新进展[J].计算机科学,2014,41(12):1-7.
[2]李瑞峰,王亮亮,王坷.人体动作行为识别研究综述[J].模式识别与人工智能,2014,27(1):35-48.
[3]徐勤军,吴镇扬.视频序列中的行为识别研究进展[J].电子测量与仪器学报,2014,28(4):343-350.
[4]秦娜,桑凤娟.基于自适应邻域选择的局部判别投影算法[J].计算机工程,2013,39(4):194-198.
[5]谢钧,刘剑.一种新的局部判别投影方法[J].计算机学报,2011,34(11):2243-2249.
[6]文贵华,江丽君,文军.邻域参数动态变化的局部线性嵌入[J].软件学报,2008,19(7):1666-1673.
[7]高玮军,白万荣,公维军,等.基于自适应邻域选择的正交局部敏感判别分析[J].计算机工程与设计,2012,33(5):1968-1972.
[8]朱旭东,刘志镜.基于主题隐马尔科夫模型的人体异常行为识别[J].计算机科学,2012,39(3):251-255,275.
[9]田国会,尹建芹,韩旭,等.一种基于关节点信息的人体行为识别新方法[J].机器人,2014,36(3):285-292.
[10]申晓霞,张桦,高赞,等.一种鲁棒的基于深度数据的行为识别算法[J].光电子激光,2013,24(8):1613-1618.
[11]金一,王移芝,阮秋琦.张量正交局部敏感判别式分析及其在人脸识别中的应用[J].信号处理,2011,27(6):820-827.
[12]张云龙,李萍,张善文.基于正交判别局部保持映射的步态识别方法[J].计算机测量与控制,2014,22(5):1644-1646.
[2]李瑞峰,王亮亮,王坷.人体动作行为识别研究综述[J].模式识别与人工智能,2014,27(1):35-48.
[3]徐勤军,吴镇扬.视频序列中的行为识别研究进展[J].电子测量与仪器学报,2014,28(4):343-350.
[4]秦娜,桑凤娟.基于自适应邻域选择的局部判别投影算法[J].计算机工程,2013,39(4):194-198.
[5]谢钧,刘剑.一种新的局部判别投影方法[J].计算机学报,2011,34(11):2243-2249.
[6]文贵华,江丽君,文军.邻域参数动态变化的局部线性嵌入[J].软件学报,2008,19(7):1666-1673.
[7]高玮军,白万荣,公维军,等.基于自适应邻域选择的正交局部敏感判别分析[J].计算机工程与设计,2012,33(5):1968-1972.
[8]朱旭东,刘志镜.基于主题隐马尔科夫模型的人体异常行为识别[J].计算机科学,2012,39(3):251-255,275.
[9]田国会,尹建芹,韩旭,等.一种基于关节点信息的人体行为识别新方法[J].机器人,2014,36(3):285-292.
[10]申晓霞,张桦,高赞,等.一种鲁棒的基于深度数据的行为识别算法[J].光电子激光,2013,24(8):1613-1618.
[11]金一,王移芝,阮秋琦.张量正交局部敏感判别式分析及其在人脸识别中的应用[J].信号处理,2011,27(6):820-827.
[12]张云龙,李萍,张善文.基于正交判别局部保持映射的步态识别方法[J].计算机测量与控制,2014,22(5):1644-1646.