An Automatic Recognition Procedure is provided to recognize military objects on simplified background，such as aeroplanes and tanks，etc. There are several steps in the digital processing procedure，they are noise filtering，object segmentation，feature extraction，classification and displacement measuring. Simple and effective algorithms are used in each step. They are median filtering for noise cleaning，histogram fitting algorithm for object segmentation，advanced moment invariant for feature extraction，nearest neighbor rule for object recognition and projection center measuring for displacement measurement of the object center. Simulation is applied to each procedure through an experimental image processing system and this method is proved to work effectively On simplified background，ie. low noise condition. Since all the algorithms are simple，the system may be accomplished with multi microprocessor cascade.