A CAPTCHA Recognition Algorithm Based on Character Segmentation and Random Forest
Suke Kong, Yu-Chung Chang · Journal of Physics Conference Series · 2023
Abstract In this paper, the CAPTCHA code is recognized based on character segmentation and random forest algorithm. In the recognition process, the gray conversion method, binarization method, and Gaussian filtering method are used to process the image, and then the projection segmentation method is used to segment the characters. Using the random forest algorithm to recognize a single character, the recognition accuracy reaches 94%. In character recognition, it is found that 5 and S, 8 and B, 4 and A, and T and F have the lowest resolution accuracy ratio.