Recognition of distorted target based on Mexican Hat Optimum Trade-off Maximum Average Correlation Height algorithm

摘要

In order to solve the problem of inaccurate recognition for distorted target (the targets rotated in plane or scale changed) in cluttered background among the image pattern recognition, combined with the Mexican Hat mother wavelet and the Optimum Trade-off Maximum Average Correlation Height (OT-MACH) algorithm, the Mexican Hat Optimum Trade-off Maximum Average Correlation Height (Mexican Hat OT-MACH) matched filter is designed. The Mexican Hat OT-MACH filter is obtained to recognize distorted target in cluttered background. The wavelet functions have the multi-scale characteristic and can analyze the specific frequency information. Moreover, the Optimum Trade-off Maximum Average Correlation Height algorithm (OT-MACH) has three characteristics, namely high distortion tolerance, suppressing noise and sharpening the correlation peak. Therefore, the new designed matched filter contains all the characteristics of the wavelet function and OT-MACH algorithm. In order to balance all the performances of the new designed Mexican Hat OT-MACH matched filter, the performance control parameters and the scale coefficient of the Mexican Hat OT-MACH matched filter need to be set. Thus, the new designed Mexican Hat OT-MACH matched filter has high versatility. It can respond higher correlation peaks and has higher distortion tolerance to recognize various types of distorted targets in cluttered background. In order to prove the feasibility of the Mexican Hat OT-MACE filter and determine its distortion tolerance, a lot of computer simulation experiments have be done with the filter. Good effect can be obtained.