On-orbit Computational Imaging Method

Abstract

The invention provides an on-orbit computational imaging method, comprising: learning a hyperspectral feature dictionary model on a hyperspectral training image set marked with ground object types to obtain a hyperspectral classification dictionary; The object types in the hyperspectral classification dictionary are subjected to abnormal target detection on hyperspectral images and infrared images to be detected and visual saliency detection on multispectral images to obtain target area data; panchromatic images and multispectral images are generated on the track The fused image corresponding to the target area data is indexed and stored to obtain compressed imaging data. The method of the invention can improve the timeliness of data utilization and reduce the pressure of transmission and storage of remote sensing images.

Description

technical field [0001] The invention belongs to the field of remote sensing imaging, relates to technical fields such as on-orbit image processing, target detection, target recognition, data compression, and particularly relates to an on-orbit computational imaging method. Background technique [0002] Each type of sensor tends to capture only one or a few features. However, each type of target has multiple types of features, such as appearance, structure, temperature, material, and category. Multiple features must be used comprehensively to identify targets accurately and comprehensively. For this reason, the same satellite is often equipped with different types of sensors, such as visible light, infrared and hyperspectral, while using multi-source images acquired by multiple sensors to detect and identify targets. [0003] The imaging process of traditional visible light, infrared, hyperspectral and other sensors is physical imaging. Physical imaging uses components to ma...

AI Technical Summary

Problems solved by technology

[0005] Computational imaging has many advantages, but computational imaging involves many steps such as target detection, data fusion, and data compression; computational imaging needs to be completed on-orbit on the satellite, with limited computing resources and high requirements for computational timeliness. The key technologies of computational imaging are urgently needed. In-depth research is required

Images