AN ALGORITHM FOR SEGMENTATION OF MERGED CHARACTERS BASED ON FOREPART PREDICTION AND RECOGNITION

Segmentation of merged characters is one of difficulties that have attracted a great deal of attention in optical character recognition （OCR）. Nowadays, unsuitable segmentation is the primary cause of recognition errors. In this paper, after analyzing several representative related approaches, an algorithm for segmentation and recognition of merged characters based on prediction and recognition is presented. At first, the first, character is predicted and cut from the image with a kind of shielding code, and then examined by the classifier. Finally, all of the characters are recognized one by one. This approach is effective and feasible, and also robust while cutting complicated merging characters even with bad quality of image.