Modified Word Recognition System--New Performance

This paper presents a hybrid system for word recognition. It is based on a modification to the view-based approach presented in authors´ previous works. The algorithm is based on recognizing the whole words without separating them into letters. The characteristic vectors taken from both top and bottom views of the image are processed with the method of Toeplitz matrices. The obtained series of minimal eigenvalues is used for classification. The procedure of classification involves arrays of neural networks. Each network in the array is responsible for detecting one class. The results of the experiments on a set of common English words are presented.