Optical Correlation Recognition Technology of Small Moving Target Based on Wavelet Multi-Scale Edge Fusion

Abstract

Moving target recognition has a wide application in varieties of domains and has received more attention in recent years. Among all the recognition methods, optical correlation technology, using hybrid optoelectronic joint transform correlator (HOJTC), has the merits of parallel, large capacity and high-speed, so it is considered as one of the most effective methods for moving target detection. But in the course of detecting, some targets are too small to be recognized by HOJTC. These small targets in cluttered background and strong noise may affect the brightness of the correlation peaks and have a serious influence on target recognition results. In order to solve this problem, this paper applies a wavelet multi-scale edge fusion algorithm which firstly extracts the edge at different scales by wavelet modulus maximum method, and then fuses the detected edge to obtain more edge information of the target. In addition, for the moving distortion problems, taking temporal state of the target as the template can achieve the template update. To prove this method, many detection experiments of small moving targets have been performed with optical correlation method. As an example a small target “boat” taking up only pixels is presented. The recognition results show that the brightness of the correlation peaks is enhanced and the target recognition ratio is increased. The conclusion can be drawn that applying this algorithm in optical correlation method can realize the small moving target recognition successfully and expand the application to scope of optical correlation technology.

Scale Distorted Target Recognition Based on Machine Vision

“…First of all, the edge features of detected target need to be extracted. In this paper, the edge extraction algorithm based on lifting wavelet [5] is adopted to process the reference image of the detected target. To describe the effectiveness of the improved MACH filter, a group of experiments for an fighter target in the sky are implemented in this paper.…”

Research on 2F Optical Correlator Based on Neural Network Filter for Recognizing Large-Angle Rotation Distortion Target

“…Compared with traditional computer pattern recognition, optical correlation recognition has advantages in ultra-high speed operation, high bandwidth, anti-electromagnetic interference and strong stability, so it has broad application prospects. At present, optical correlation recognition has been applied to complex background moving target recognition, face recognition and other fields, playing its unique advantages [1]- [4]. The matched filter in the optical correlator is the key device to determine the recognition performance of the system.…”