In this research work we propose a fast algorithm for automatically recognizing a limited set of gestures from hand images for Zig-Bee interfaced device control application. Hand gesture recognition is a challenging problem in its general form. We consider a fixed set of manual commands and a reasonably structured environment, and develop a simple, yet effective, procedure for gesture recognition. Our approach contains steps for segmenting the hand region, locating the fingers, and finally classifying the gesture. The algorithm is invariant to translation, rotation, and scale of the hand. As a part of work we also analysis the mean deviation approach with fast algorithm. From the recorded Results of hardware & software we demonstrate the effectiveness of the fast algorithm on the basis of time & accuracies. Thus the system we describe here may be the "Soft Remote Control System" which can provide easy control of home installed appliances to improve the inhabitant's comfort.