An Example-Based Method of Touching String Segmentation

摘要

In the work, an example-based method of touching string segmentation is proposed. Using Markov random field, the candidate patches based on the compatibility of the neighbour patches are selected. The outputs of the MRF after the iterative belief propagation form a segmentation probability map. The cut position is extracted from the map. Experiment results are presented and demonstrate the effectiveness of our algorithm.