VEHICLE LOGO RECOGNITION USING ENHANCED LEARNING : VLR Recognition

VLR enrolls a vital role in vehicle identification for the intelligent traffic systems. Recognition approach typically gives much importance to the training dataset. In this paper an efficient algorithm is proposed using EUCLIDEAN DISTANCE, COLOR HISTOGRAM USING K-MEANS, HOG and KNN, to extract features of vehicle logos and to classify each brand of logos. Here training dataset is shaped by applying Euclidean distance and k-means to the raw dataset. Testing data samples were also taken from the standardized data sets. At the final stage, accuracy of the logo classification is calculated based on CCA and MCCA. CCA is class-wise accuracy, whereas MCCA is overall accuracy for VLR system. This algorithm is best suited for classifying complex and similar logos of different vehicles brands.