The paper proposes a system for gesture or hand images recognition to provide a easy method for communication between deaf-dump and vocal people it allow the user to deal with the system in a natural way. The hand can be classified into one of the four gesture classes or one of the four different movement directions. The aim of this paper to develop a system which converts the 32 combination of binary number 25 into decimals and then the decimals number converted into a sign language. In this paper we develop a system by using feature extraction method, HMM ( Hidden Markov Model ), canny edge detector and image processing tool.