Abstract: When observation distance and angle change, angular distortion and scale distortion of target image relative to the template will appear, which limits the development of pattern recognition with correlation recognition technology. The Maximum Average Correlation Height (MACH) filter was applied to recognize distorted targets in this paper. According to multiple computer simulation experiments and optical experiments, the filter has the characteristcs of high distortion tolerance and bright correlation peaks by optimizing the control parameters of the synthesized filter. Distorted targets were filtered in frequency domain by the improved MACH filter, correlation peaks could be sharpened effectively and recognition scope was expanded succesfully. As a practical example, computer simulation experiments and optical experiments for warship in cluttered scene were carried out. The experimental results prove the feasibility of the algorithm.