Thai Traffic Sign Classification and Recognition System Based on Histogram of Gradients, Color Layout Descriptor, and Normalized Correlation Coefficient

摘要

The traffic sign is one of the useful features for improving driving assistance technology. There are some different patterns in each country. As a result, this paper proposes the classification and recognition system, which are applicable to the Thai traffic sign. In the classification process, two classifiers: support vector machine (SVM) and random forest are combined and run on two selected features: histogram of oriented gradients (HOG) and color layout descriptor (CLD) to categorize an incoming sign into one of four classes. In this process, the proposed technique provides the accuracy score up to 93.98%. Subsequently, the recognition process is used to identify each traffic sign. This process is based on the OCR and correlation coefficient-based template matching. From the experiments, our proposed system can be applied to a sign image and used as a part of driving assistance system.