A Method for Locating and Segmenting Character Blocks of Print Form Using Active Template

摘要

Form is an important information carrier; the recognition of form is becoming more and more important with the development of optical character recognition (OCR). Because of the floating of character blocks in print forms, it is one of the most important steps to locate character blocks and segment these blocks into individual characters. In this paper, we propose a new method for locating character blocks and segmenting them into characters. First, an active template structure is defined to describe the character blocks of the forms. Each item in the structure represents one of the character blocks to be recognized or their relative location. Then character blocks are located by using the active template. Finally, character blocks are de-skewed and character segmentation is fulfilled in each character blocks. Experimental results show that our method is very robust for character location and segmentation of print forms.