Sign language is means of communication. Some of them are not ever learn to sign language. This increases the isolation of deaf people and they may be confined in many of their express to communicate only with other deaf people. Preprocessing, feature extraction and classification are three issues in sign language recognition system. This work presents a method to improve the sign language recognition rate using Hough transform, Wavelet Transform artificial neural network. Hough transform and Wavelet Transform are applied to obtain features. Feed Forward Network trained by back propagation algorithm are developed to recognize sign images. Experimental results how that effectiveness of the proposed approach for sign language recognition with good recognition accuracy.