Construction Research on Multi-Threshold Segmentation Based on Improved Otsu Threshold Method

The aim of unstructured road detection is the recognition of road region and the tracking of road boundary. An improved Otsu threshold method available in double-threshold segmentation was proposed in this paper to resolve the multi-peaks problem in both complicated background and target. This method remedies the mis-segmentation of Otsu threshold method caused by huge diversity of between-class variances and refines the segmentation precision by matching the proposed Otsu edges with the weighted Canny edges. According to the irregular characteristics of the unstructured road and the requirement of the real-time motion planning for intelligent vehicle, improved Otsu threshold method was proposed. The experiment results in different road environments demonstrated the validity and real-time of the proposed method.