Abstract: The basic features are high resolution range profiles for target recognition in high frequency region The data base should contain all possible aspects of a target But in real case,the data base couldn't be too large So the measured data(profiles)should be compressed first The max min correlation template was applied to the data base It operated on profiles within a range of attitude All of these profiles were substituted by one template Results show that the max min correlation template is better than the average template in the respect of recognition efficiency