基于分层梯度方向直方图和SVM的人体识别.PDFVIP

. 2010. 18( 11) 2602 Computer Measurement Control : 2010) : T P391 : A SVM 李 晖, 曲仕茹 ( , 710072) : , ( H OG) ; ( PCA) , HOG+ PCA ; ( RBF) ( SVM ) , , ; , H OG : ; ; ; ; Human Recognition Based on Hierarchical Histograms of Oriented Gradients and Support Vector Machines i H ui, Qu Shiri ( College of Automation, Northwestern Polytechnical University, Xi an 710072, China) Abstract: In order to further improve the precision of human r ecognition, this paper pr esents a new feature called w eighted hierarchical histogram s of oriented gradients ( H OG) , which can reflect the overall shape inform ation of the local image. Principal component analysis ( PCA) algorithm is applied to improve the structure of the proposed featur e, and g et another new feature called hierarchical H OG+ PCA; Support vector machines ( SVM) based on radial basis function ( RBF) kernel is selected as human classifier, and its best parameters are got by the cross- validation procedure. T he ex perim ents conducted on the complete training and test sample set show that both of the recognition precisions of the proposed features outperform the ones of the HOG. Ke words: human recognition; hierarchical histograms of oriented gradients; principal component analysis ( PCA) ; support vector ma chines ( SVM) ; parameter optim ization 0 , , Dalal Triggs[4] , SIFT ( H istograms of Oriented Gradients, H OG) , HOG ( , , ) ,