Research about text image segmentation based on the algorithm of LLT

Character segmentation is the area of image segmentation, which is the classical problem. Different researchers put forward different resolution algorithms, but mainly concentrate on whether the segmentation result is good or not. As for the distinct difference between the prospect and background of image, global threshold segmentation method is simplest, but it is not suitable for the image which has noise and uneven illumination. Global threshold segmentation method which is sensitive to noise can solve the problem of uneven illumination, but the divided character stroke may be uneven and a speckle noise on local window can be easily divided into the prospect, which can be dealt by the well-adapted and fast algorithm of LLT. This passage introduces the algorithm of LLT, and on its basis analyzes character segmentation problem, then raising a new improved method to calculate threshold and getting a satisfactory result.