Vehicle Target Recognition Algorithm based on Multidimensional Feature Fusion and Adaboost-SVM Strong Classifier
Download as PDF
Two image preprocessing methods have been introduced, comparative analysis has been made with combination of experiments; three traffic sign detecting methods have been summarized, including detecting algorithms based on color, shape and both; experimental comparative analysis has been made for the extraction methods for various traffic sign features, the experiment has proven that the probabilities of wrong identification for triangle and round marks are highest; after studying the features of existing traffic sign recognition method AdaBoost and SVM, One changing AsaBoost technology, traffic sign detecting method combining color and shape, feature extraction method of sub-pattern combination has been adopted. On the basis of sub-model, it compares adjacent block, overlapped edge block and slide block method and distinguishes common traffic signs with the recognition method of support vector machines classifier combination based on radial basis function.
Vehicle Target Recognition, Classifier, Image, Feature Fusion.