Research on Automatic Sampling for Handwritten Characters
Liguo Deng · Sichuan University of Science and Technology · 2004
A method to code character image based on information of run length is presented, which can save a mass of file spaces with binary form to access data. In preprocessing, optimal threshold value method and smooth technology are adopted to process sample form image. In order to segment character image correctly, vertical edge of sample form image is extracted firstly, then Hough transform is used to detect slant angle of image. Also, a fast iteration method based on center of run-length for character image segmentation is proposed. Testing results of handwritten numerals sampling show that effectiveness of the method.