Abstract:Vehicle-license-plate character segmentation is one of the most important steps in license plate recognition. Traditional methods always need high quality original images and thus, will inevitably affect the ultimate recognition rate. In this paper, a method for segmentation of license plate characters, based on Renyi’s entropy and mathematical morphology edge detection, is introduced. The algorithm binaries the license plate by two-dimensional Renyi’s entropy first and then detects the edges by morphology erosion. After getting rid of the outer bounding frame, the license plate character is segmented by projection. Computer simulation results show that the character is much clearer after preprocessing based on Renyi’s entropy and mathematical morphology edge detection. This is favorable for the character segmentation based on projection. The algorithm is not only simple and fast, but also robust, and performs better than traditional methods.