The present character recognition system is developed for an isolated character, and it need a segmentation process as preprocessing for the image of the touching character string. The process segments them into some isolated characters. To improve the total performance of the character recognition system, it is required to lessen the number of retrieval path and the number of the isolated character to recognize in the candidate lattice as much as possible. And it is required that the segmentation process don't arise the distortion in the isolated characters to recognize correctly. We proposed the segmentation method based on connecting condition of neighbor lines to resolve these problems. In this paper, we propose a new segmentation algorithm, which develops our previous method, and we evaluate the performance of these segmentation methods by the direct evaluation system with redundancy degree and the indirect evaluation using the character recognition. We confirm the usefulness of our new method by the comparison of three segmentation methods.
展开▼
