Extrating Road Direction Information from Road Sign Imagery Taken by Movile Mapping System

Since road signs are officially managed by local government, they are reliable information for drivers. For systematic managing the road sign information, it is necessary to construct road sign database. In this paper, we deals with automatic road direction information extracting algorithm from road sign imagery obtained by MMS (Mobile Mapping System). Our algorithm consists of three phase. In the first phase, we convert the gray-scale image to binary image with Otsu algorithm to distinguish the objects (arrow, character, symbol etc.) from the background. This phase is followed by automatic extracting arrow region with the proposed algorithm. Proposed algorithm is finding the seeds of arrow region by increasing the size of search window. After finding the seeds, region growing method is applied to extract the arrow region. In the final phase, we recognize the direction information from the arrow by LSM (Least Square Matching) algorithm. From the arrow area, all the corner points are detected. Not all the corner points comprise the arrow heads. Hence, LSM (Least Square Matching) algorithm is applied to the corner points to extract the direction information. For the LSM algorithm, eight direction arrow head shaped templates are used. Proposed algorithms are applied to 100 test images, which are obtained by MMS.