Traffic Sign Detection and Recognition for Assistive Driving

The technology behind Advanced Driver Assistance Systems has been continuously advancing in recent years. This has been made possible by artificial intelligence and computer vision. In Automatic Traffic Sign Detection and Recognition System, accurate detection and recognition of traffic signs from the complex traffic environment and varying weather and lighting conditions are still a big challenge. This study implements a traffic sign detection and recognition system. Bilateral filtering pre-processing technique is performed before detection phase to improve accuracy. Color thresholding in HSV color space followed by Hough transform are used for a more efficient segmentation of the region of interest. In recognition phase, Histogram of Oriented Gradients is extracted from candidate traffic signs as the key feature in classification. This study also determines which machine learning classifier will provide the best accuracy for traffic sign recognition. The classifiers evaluated are K Nearest Neighbor, Support Vector Machine, Gaussian Process, Decision Tree, Random Forest, Multilayer Perceptron, AdaBoost, Gaussian Naive Bayes, and Quadratic Discriminant Analysis. This study has determined that bilateral filtering provides improvement in accuracy with 2.02% more in detection, 0.68% less in non-detection and 1.35% less in false detection. Detection accuracy is at 68.25% for dataset from online sources and an effective accuracy of 75% for local traffic images. Multilayer Perceptron Classifier obtained the highest accuracy (0.9), precision (0.9), recall (0.9) and f1 score (0.91) for traffic sign recognition.