计算机工程与应用
計算機工程與應用
계산궤공정여응용
COMPUTER ENGINEERING AND APPLICATIONS
2013年
8期
156-159
,共4页
人体行为识别%神经网络%Canny算子%不变矩特征
人體行為識彆%神經網絡%Canny算子%不變矩特徵
인체행위식별%신경망락%Canny산자%불변구특정
human action recognition%neural network%Canny operator algorithm%moment invariant feature
为了提高了人体行为识别的正确率,提出了一种基于改进Canny算子和神经网络的人体行为识别模型(ICanny-RBF).采用改进Canny算子对人体行为图像进行预处理,提取人体行为轮廓,提取7个不变矩特征作为RBF神经网络的输入向量,训练出能够识别人体行为的RBF神经网络模型,并采用取k-means算法确定RBF神经网络聚类中心,采用Weizmann数据集进行仿真实验.仿真结果表明,与传统方法相比,提出的ICanny-RBF模型提高了人体行为的识别正确率.
為瞭提高瞭人體行為識彆的正確率,提齣瞭一種基于改進Canny算子和神經網絡的人體行為識彆模型(ICanny-RBF).採用改進Canny算子對人體行為圖像進行預處理,提取人體行為輪廓,提取7箇不變矩特徵作為RBF神經網絡的輸入嚮量,訓練齣能夠識彆人體行為的RBF神經網絡模型,併採用取k-means算法確定RBF神經網絡聚類中心,採用Weizmann數據集進行倣真實驗.倣真結果錶明,與傳統方法相比,提齣的ICanny-RBF模型提高瞭人體行為的識彆正確率.
위료제고료인체행위식별적정학솔,제출료일충기우개진Canny산자화신경망락적인체행위식별모형(ICanny-RBF).채용개진Canny산자대인체행위도상진행예처리,제취인체행위륜곽,제취7개불변구특정작위RBF신경망락적수입향량,훈련출능구식별인체행위적RBF신경망락모형,병채용취k-means산법학정RBF신경망락취류중심,채용Weizmann수거집진행방진실험.방진결과표명,여전통방법상비,제출적ICanny-RBF모형제고료인체행위적식별정학솔.
In order to improve the correct rate of recognition of human action, a human action recognition model is proposed based on improved Canny operator and neural networks in this paper. Improved Canny operator is used to preprocess the human action image to extract the contour of human action image, and then the 7 HU moment invariant features are extracted as input vector of RBF neural network and it trains to build RBF neural network model for the human action, and uses k-means algorithm to determine RBF neural network clustering center, simulation experiments are carried out on Weizmann data set. The simula-tion results show that the proposed model improves human action recognition correct rate and it is an efficient, high correct iden-tification human activity recognition method.