Published Paper Details:

Internet registration usually requires one sort of the Completely Automated Public Turing test to inform Computer and Human Apart (CAPTCHA) to avoid illegal registration through computer programs called bots. The CAPTCHA script is formed of distorted or noised characters. This paper presents an analytical study of extraction of varied statistical features for recognition of every character in CAPTCHA image. OCR (Optical Character Recognition) is generic process in area of pattern recognition and is aimed to recognize each character in CAPTCHA image. OCR system involves basic four stages: preprocessing, segmentation, feature extraction, classification which is followed by actual recognition. The technique of feature extraction plays an important role in character recognition. The CAPTCHA character recognition depends on the extracted features. The various statistical feature extraction techniques have been applied over character image. Here, the various types of statistical feature extraction methods are discussed for the application of character recognition for CAPTCHA verification.