Feature extraction in character recognition with a neural network

Neural net is capable to recognize handwritten characters and the keg of is the extraction of features if the pattern feature do not include enough information of feature recognition objects or cannot extract the structure information reflecting the object feature, they can not be recognized therefore, the rapid and effective extraction of the features which reflect the structure information of objects is the key of pattern recognition due to joining script and separate script, handwritten characters are the problem of segmentation. Segmentation is the initial problem of recognition this paper purposes the adoption of measures, such as fuzzy set theory membership function, improved character segmentation the feature extraction of uneven net density, to accomplish the segmentation the feature extraction of uneven net density, to accomplish the segmentation of characters and the feature extraction, thus to increase the recognition rate of characters.