Indonesian traffic sign detection and recognition using color and texture feature extraction and SVM classifier

This paper presents traffic sign detection and recognition which is necessary to be developed to support several expert systems such as driver assistance and autonomous driving system. This study focused on the detection and recognition process tested on Indonesian traffic signs. There were some major issues on detecting process such as damaged signs, faded color, and natural condition. Therefore, this paper is proposed to address some of these issues and will be done in two main processes. The first one is traffic sign detection which divided into two steps. Start with segmenting image based on RGBN (Normalized RGB), then detects traffic signs by processing blobs that have been extracted by the previous process. The second process is traffic sign recognition process. In this process there are two steps to take. The first one is feature extraction, in this research we propose the combination of some feature extraction that is HOG, Gabor, LBP and use HSV color space. In next recognition stage some classifier are compared such as SVM, KNN, Random Forest, and Naïve Bayes. The propose method has been tasted on Indonesia local traffic sign. The results of the experimental work reveal that the approach of RGBN method showed precision and recall about 98,7% and 95,1% respectively in detecting traffic signs, and 100% for the precision and 86,7% for recall in recognizing process using SVM Classifier.