Fuzzy C-Mean: A Statistical Feature Classification of Text and Image Segmentation Method

Classification of text and image using statistical features (mean and standard deviation of pixel color values) is found to be a simple yet powerful method for text and image segmentation. The features constitute a systematic structure that segregates one from another. We identified this segregation in the form of class clustering by means of Fuzzy C-Mean method, which determined each cluster location using maximum membership defuzzification and neighborhood smoothing techniques. The method can then be applied to classify text, image, and background areas in optical character recognition (OCR) application for elaborated open document systems.