OPTICAL correlation identification technology applied in underwater laser imaging target identification

Abstract

The underwater laser imaging detection is an effective method of detecting short distance target underwater as an important complement of sonar detection. With the development of underwater laser imaging technology and underwater vehicle technology, the underwater automatic target identification has gotten more and more attention, and is a research difficulty in the area of underwater optical imaging information processing. Today, underwater automatic target identification based on optical imaging is usually realized with the method of digital circuit software programming. The algorithm realization and control of this method is very flexible. However, the optical imaging information is 2D image even 3D image, the amount of imaging processing information is abundant, so the electronic hardware with pure digital algorithm will need long identification time and is hard to meet the demands of real-time identification. If adopt computer parallel processing, the identification speed can be improved, but it will increase complexity, size and power consumption. This paper attempts to apply optical correlation identification technology to realize underwater automatic target identification. The optics correlation identification technology utilizes the Fourier transform characteristic of Fourier lens which can accomplish Fourier transform of image information in the level of nanosecond, and optical space interconnection calculation has the features of parallel, high speed, large capacity and high resolution, combines the flexibility of calculation and control of digital circuit method to realize optoelectronic hybrid identification mode. We reduce theoretical formulation of correlation identification and analyze the principle of optical correlation identification, and write MATLAB simulation program. We adopt single frame image obtained in underwater range gating laser imaging to identify, and through identifying and locating the different positions of target, we can improve the speed and orientation efficiency of target identification effectively, and validate the feasibility of this method primarily.