Remote sensing target intelligent detection and recognition system based on optical deep neural network

Abstract

A remote sensing target intelligent detection and recognition system based on an optical deep neural network is characterized in that a remote sensing image light field is regulated and controlled to equivalently replace the calculation function of a traditional electronic device, so that an optical system directly has the functions of processing, classifying and recognizing images by the deep neural network; the all-optical deep neural network intelligent remote sensing detection technology works in the form of light, and has the outstanding characteristics of all-optical calculation, extremely simple hardware implementation, instantaneous processing of mass data, extremely low power consumption and the like; the all-optical deep neural network enables the intelligent remote sensing load system to use all-optical calculation to replace a traditional electronic device on the aspect of the overall design thought, thereby greatly reducing the dead weight of the remote sensing system, and essentially solving the dependence of remote sensing load intelligent target recognition on a huge amount of calculation resources. And a feasible technical route is provided for realizing on-orbit intellectualization of the remote sensing system.

Description

technical field [0001] The invention relates to an intelligent detection and recognition system for remote sensing targets based on an optical deep neural network, and belongs to the technical fields of all-optical computing, deep learning and optical remote sensing. Background technique [0002] As the most active research direction in today's information science, artificial intelligence has important applications in the aerospace field. Artificial neural network, as the most important model of artificial intelligence, is widely used in various scenarios because of its good generalization ability and robustness. The artificial neural network establishes the connection between neurons in each layer of the neural network by imitating the structure of the nervous system. Integrated circuit chips are the hardware carrier for training and testing of today's mainstream neural network models. Traditional neural networks can run on CPUs, GPUs, FPGAs, and application-specific integ...

AI Technical Summary

Problems solved by technology

[0005] The technical problem solved by the present invention is: Aiming at the lack of feasible technical approaches of the all-optical deep neural network for directly intelligently processing incoherent natural light remote sensing images in the current prior art, a remote sensing target intelligence based on optical deep neural network is proposed. Detection and identification system

Images