An EMU Character Image Segmentation Algorithm Based on Sobel Operator

• 1、Automation School, Huazhong University of Science and Technology,Wuhan 430074

Abstract：Aiming at the automatic recognition of EMU license characters in TEDS, this paper presents a method of EMU characters image segmentation based on Sobel. Firstly, extract edge information in the image using Sobel operator, initially location the background region and the target region in the image by the edge information of the image, and then set the sliding window analysis the edge information in the window, according to the edge information segment the image. This method considers the EMU characters images are acquired by linear array scanning, makes the law of illumination has a certain direction, so using the Sobel operator to extract edge information in the illumination direction and initial positioning. Relative to the global threshold segmentation algorithm and the local threshold segmentation algorithm, this approach is less affected by the illumination. Repeated experiments showed that this method in uneven illumination conditions has a better effect for the EMU characters image segmentation, and provide better image conditions for character recognition.

Keywords： Image Processing Edge Information Uneven Illumination Image Segmentation TEDS