Feature Extraction Methods for Handwritten Character Recognition
Optical character recognition is the newest technology in the last decade in the area of image processing, pattern recognition and machine learning. There is very much scope to develop a machine to recognize handwritten English character efficiently, because of writing styles of everyone different in size, measurement, slats etc. The problem in recognition is focused with feature extraction methods and classifiers. This paper addressed for extraction of features with help of blind source separation process as independent component analysis. The Particle swarm optimization (PSO) and firefly algorithms (FFA) are applied separately for instant selection process. Due to distributed neighborhood pixel of an image, the PSO gives better recognition rates, but take large time for execution, whereas FFA has less recognition rate but it used high speed of selection. It is part of the Content Based Image Retrieval (CBIR) program that solves the delinquent image search in huge data collection. Particle swarm optimization and firefly algorithms are impl0.emented for feature vector selection. It is observed that the PSO gives better recognition levels due to scattered neighborhood pixels of an image.