Dual-channel satellite attitude estimation network based on deep learning

Abstract

The invention discloses a dual-channel satellite attitude estimation network based on deep learning. The network process comprises the steps of constructing a satellite attitude estimation data set and dividing the satellite attitude estimation data set, preprocessing data, extracting image features by a ResNet model, learning spatial position information by using an improved Vision Transformer model, learning rotation information by using a Hurglass network, respectively outputting position information and attitude information of a satellite, calculating an output value and a label distance, and outputting the output value and the label distance. And carrying out back propagation on the model to carry out iterative training to obtain an optimal model, and evaluating the model by using a test set. A self-attention mechanism of a core module of the Vision Transform model is improved, so that the position of a satellite in an image can be focused, and key points of a satellite contour can be effectively deduced through repeated up-sampling and down-sampling processing means of Hurglass network learning. According to the method, the spatial position information and the attitude information of the satellite are respectively learned by using the two-channel network, so that the mutual influence between the two kinds of information is effectively avoided, the satellite attitude estimation accuracy is improved, and a new intelligent means is provided for spatial non-cooperative target detection.

Description

technical field [0001] The invention relates to the field of target detection, in particular to a dual-channel satellite attitude estimation network based on deep learning. Background technique [0002] Vision-based satellite attitude estimation is an urgent problem to be solved in the aerospace field, and it has extremely important application value in navigation, on-orbit maintenance and space junk cleaning. However, satellite attitude estimation based on pure vision faces many technical problems that need to be solved, such as the inconvenience caused by the space illumination environment for camera imaging, and the interaction between the satellite's spatial position information and its own rotation information. The solution for satellite attitude estimation in the world presents many challenges. [0003] Vision Transformer simulates the attention mechanism of the human brain. When learning image features, it can automatically focus on the detected object. This method c...

AI Technical Summary

Problems solved by technology

[0004] The invention patent of the authorized publication number CN 109827578 B discloses "the method of satellite relative attitude estimation based on contour similarity", which uses the simulation image for projection to analyze the similarity of the satellite image contour, but the method is not effective in recognition efficiency and robustness. lacking in

The invention patent with the authorized publication number CN 105300384 B discloses "an interactive filtering method for satellite attitude determination". By collecting relevant data from satellite sensors, the satellite attitude information is determined. This method is only used for the satellite body to determine its own attitude. measurement, unable to identify non-cooperative satellite attitude information

Images