Flying target recognition system based on photon neural network and construction method

Abstract

The invention discloses a flight target recognition system based on a photon neural network and a building method, and aims at high-speed motion aircrafts, including but not limited to aircrafts such as aircrafts, unmanned aerial vehicles, helicopters, gliders and delta wings, missiles and the like, and an ultrafast flight target recognition system of the photon neural network is built. The invention provides a flight target recognition system based on a photon neural network. The flight target recognition system based on the photon neural network comprises an imaging and filtering module, a light field scaling and coupling module, a photon neural network module, a mode regulation and control module and an output light field coupling module. The invention discloses a building method of a flight target recognition system. The building method comprises the steps of artificial neural network design and artificial neural network photonics implementation. According to the invention, a photon neural network transmission and regulation model based on all-optical passive components is designed, training is carried out for small sample learning of a high-speed flight target, a static and dynamic target identification system is established, and identification of a high-speed flight moving target is realized.

Description

technical field [0001] The invention relates to the field of artificial intelligence, relates to machine vision and target recognition technology, and more particularly relates to a photon neural network-based flight target recognition system and construction method, which can be used in industrial production, modern logistics, disaster warning and national security. Background technique [0002] Target identification technology (target identification technique) is a technology that uses photoelectric detection equipment and computer to identify distant targets, uses computer to extract target feature information from the signal obtained by photoelectric detection equipment, and calculates target situation estimation parameters. Finally, according to a large number of training samples The identified discriminant function is used for identification decision in the classifier. Target recognition technology recognition plays an important role in industrial production, modern lo...

AI Technical Summary

Problems solved by technology

Photonic chips have a high degree of integration, but cannot directly process optical images, and require an optical-electrical-optical conversion process, which has low efficiency and is difficult to process

The diffractive photonic neural network is simple and easy to implement, but its flexibility is low, and the system volume gradually increases with the complexity of the network structure

The scattering medium photonic neural network has a compact structure, but it is difficult to design and process

Therefore, there is still a lack of a photonic neural network target recognition system that combines the above advantages to solve the problems of low recognition speed, large volume, and high power consumption of traditional target recognition systems.

Images