Segmentation For Uneven Lighting Document Image

Image segmentation is one of the most important subjects in image processing. It seems simple, but it's a difficult matter to improve segmentation quality. In nowadays, digital technique develops rapidly, and it's easy to achieve document image in uneven lighting condition. So segmentation technique for uneven lighting image will be more and more important.This paper aims at researching segmentation technique for uneven lighting image. It uses sub window technique and proposed some methods to dispose ghost objects. The paper also uses the maximum variance between groups as information measure and improves the adaptive window selection method. Compared with Qingming Huang's method, the algorithm proposed by this paper shows improvement in speed while attests to the quality of segmentation image.Subpixel technique is a useful method in image processing. In this paper, it is introduced into image binarization. This paper proposed two simple methods to achieve subpixel precision. The experiments show that the proposed method using subpixel technique can improve segmentation quality effectively.Besides, this paper did some researches on generating chain code, extrema detection in digital images and thresholding algorithm using gradient operator.