Road sign detection method based on fast HDR image generation technique

Automatic sign recognition is an important aspect of the intelligent transportation systems. One of the challenging issues in this field is a sign detection algorithm, robust under various lighting and weather conditions. In the proposed method this problem was solved by using fast High Dynamic Range (HDR image generation technique. Three differently exposed images are used as an input instead of a single one. These images are combined together in order to recover details which could be lost due to difficult lighting conditions. To make the image generation process faster, we implemented a direct intensity mapping technique, which makes possible to obtain the combined image without recovering physical irradiance values. After getting the improved image we apply a color-based segmentation method followed by the filtering process which is based on geometrical and color properties of a signs.