A Simple and Low Cost Method for Traffic Sign Measurement Based on GPS and Epipolar Geometry

Abstract

Traffic signs are an important part of traffic signals and an important element of high-precision navigation maps, which can provide accurate positioning for autonomous vehicles. In this paper, traffic signs are measured based on vehicle monocular camera and GPS, feature point extraction and matching, and related mathematical methods. By filtering out the dynamic targets in the picture and evenly extracting the feature points in the picture region, the measurement accuracy and stability are further improved. Finally, it is determined that the picture is evenly divided into 16 regions of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$4\mathrm{X}4$</tex> . The comparison between the measured value and the actual value of the perimeter of the road sign shows that the size error of 88.5 % of the test cases is less than 10%, and the relative size error of only 4% of the test cases is more than 20 %. It provides a low-cost and efficient technical scheme for road sign mapping of high-precision navigation map.