Satellite-borne SAR real-time imaging device

Abstract

The invention discloses a satellite-borne SAR real-time imaging device, and the device adopts two imaging modes of wide-area wide-width low-resolution general survey and high-resolution detailed survey to construct SAR imaging, combines satellite-borne SAR real-time imaging processing, target detection and a target classification and recognition technology based on deep learning to realize rapid and effective reconnaissance of a sensitive area, and obtains the number, positions, types, confidence coefficients, slices and other intelligence information of interested targets; the problems of large ground processing time delay, low timeliness and the like caused by direct downloading of massive high-resolution SAR data are effectively solved, the data transmission bottleneck is effectively broken through, the solid storage and load utilization rate is improved, and the combat effectiveness of a satellite is remarkably improved; the input data of the deep learning forward reasoning model is the floating point complex data, and the floating point complex data not only contains the target feature information, but also contains the phase information of the target, so that the informationdimension is increased, the information precision is high, and the classification accuracy is improved.

Description

technical field [0001] The invention belongs to the technical field of satellite payload data processing and platform control, and in particular relates to a spaceborne SAR real-time imaging device. Background technique [0002] Spaceborne SAR has the characteristics of high spatial resolution and large coverage area, which can realize the precise positioning of ship targets, and high-resolution SAR images provide the possibility for ship type identification, but it is difficult to effectively verify the ship extraction results, making the target The performance of extraction and type identification cannot be effectively improved. The deep learning technology emerging in recent years has been widely used in object classification, recognition, video analysis and natural language processing. With the help of deep learning technology, on-orbit target detection and identification based on remote sensing images can effectively improve the use efficiency of satellites, and timely...

AI Technical Summary

Problems solved by technology

[0009] (3) The existing method mainly considers the application on the ground or the participation of ground equipment to complete the functional closed loop. It must rely on the space-ground data transmission system and occupy a large amount of satellite-ground communication bandwidth, which will inevitably consume a lot of processing time and reduce the communication efficiency of the satellite-ground network. Further affect the effectiveness of remote sensing satellites, processing efficiency and real-time will not be able to meet the needs of modern remote sensing satellites

Images