Traffic sign change detection based on grayscale adaptive SVM

Abstract

The use of vehicle image detection to detect traffic information along roads is a key technology for high-precision navigation map updating. At present, China's roads are developing rapidly. Traffic signs along the roads are constantly updated as the roads change. An accurate and efficient method for detecting traffic signs along the roads is yet to be proposed. Aiming at this problem, this paper proposes a road traffic signage change detection method based on Grayscale Adaptive SVM (Support Vector Machine). First extract the traffic signs in the two-stage image, rotate and interpolate the operation, automatically rotate the guide card to the horizontal, and zoom to the uniform size, grayscale the image pair and extract the changed part using the SVM model, and finally use the morphological operator. Process images to remove the effects of "salt and pepper" noise. Through experimental verification, this method can detect changing areas accurately and efficiently, eliminate the influence of parallax, and have a higher degree of automation.