On the study of moving objects detection and pattern recognition using LS-SVM

Based on pattern recognition theory and Support Vector Machine(SVM) technology, moving objects automatic detection, recognition and classification method are discussed in detail. An algorithm of moving objects detection on the combination of double inter-frame difference ‘or’ operating and a new multi-sorts classification method—binary exponent classification are presented. By comparing SVM with BP neural network in Vehicle classification, Experimental results showed that SVM algorithm improve recognition rate and the can recognize and classify moving objects rapidly and effectively.