Unstructured Road Detection Method Based on RGB Maximum Two-Dimensional Entropy and Fuzzy Entropy

Abstract

To solve the problem that lane keeping function for automatic driving and vehicle assisted driving will not work reliably on unstructured road without lane lines or other guide markings, this article uses the characteristics of information entropy to generate the RGB entropy image to pre-segment the road region on unstructured road image. At the same time, the maximum two-dimensional entropy algorithm is introduced to achieve the joint segmentation using gray and neighborhood gray to effectively reduce the impact of interference on segmentation. After that, the fuzzy entropy algorithm is used to judge and determine the actual road boundary by combining the results of RGB and maximum two-dimensional entropy image. Finally, using the improved least square fitting quadratic curve model to build the road boundary. Our method could well and rapidly extract the lane from unstructured road image and fit out the lane line, which helps to achieve visual based lane keeping on unstructured road for autopilot and driver assistance system.