Research On Rotation And Scaie Invariance In Target Detection And Recognition Of Optical Correlation

Optical correlation target detection and recognition with the characteristic of high parallel processing and large capability belongs to the category of optical information processing. This paper realizes the detection, recognition and location of optical image by hybrid optoelectronic real-time joint transform correlator which not only has great speed for disposing but also has high precision for detecting and identifying.However, in practice, the joint transform correlator is difficult for target recognition because the target image information which is collected from the imaging sensor has large setting noises and some variation such as rotation and scale etc. so the distortion-invariable recognition method is an important research task in the field.Based on the study Of rotation and scale invariant image recognition technology, the algorithm of maximum average correlation height (MACH) is adopted, realized and modified successfully. The modified MACH algorithm is used to auto detect and recognize in optical correlation target and to filter for joint transform power spectrum plane, which enhanced the contrast of correlation peak. The target detection accuracy and recognition ratio is improved, the range of measurement is also increased greatly and good experiment result is got. The modified MACH algorithm is the bottleneck technology of solving distortion invariant (rotation invariant and scale invariant) image recognition.