Key issues about extraction of target character in creation of IR image environment is studied. It presents a multi stage data flow that consists of three level functional modules, which are adaptive background perception and subtracting operation, selection of automatic threshold used for clutter suppression, shape analyzing and target character recognition. A morphology based data processing approach proposed in this paper is applied to background estimation. Experimental results show that the method is able to provide good filtering results and robust adaptability to image environment with clutter background.