An end-to-end impact crater detection and recognition method based on a fully convolutional neural network structure

Abstract

The invention discloses an end-to-end impact crater detection and identification method based on a fully convolutional neural network structure. The method simultaneously realizes the detection and identification of impact craters in remote sensing images of celestial bodies. The network established by the method is named CraterIDNet, and the established The network consists of two impact crater detection channels and one impact crater identification channel. The network weight parameters are only composed of convolutional layers without full connection layers. The invention proposes a candidate frame scale optimization and density adjustment mechanism for the impact crater detection channel, realizes the optimal candidate frame selection, greatly improves the detection performance of small impact crater targets, and uses different receptive fields to synchronize multi-scale impact crater targets Detection, so that the network has the ability to detect large-scale changes in impact crater targets. For the impact crater identification channel, a grid pattern layer is proposed to generate a rotation and scale invariant grid pattern map to realize impact crater identification without building a matching feature database. The invention enhances the recognition robustness.

Description

technical field [0001] The invention relates to the technical field of remote sensing image processing and astronomical autonomous navigation, in particular to an end-to-end impact crater detection and identification method based on a fully convolutional neural network structure. Background technique [0002] Impact craters are the most abundant topographic structures on the surface of celestial bodies, and their morphological characteristics and spatial distribution are important basis for the study of planetary geology. In addition, impact craters are ideal landmarks for autonomous navigation of spacecraft. The detection and identification technology of impact craters is extremely important to the study of planetary geology and the realization of autonomous navigation of spacecraft. Currently, impact crater detection and recognition are studied separately as two independent algorithms. The goal of impact crater detection is to determine whether the image contains impact ...

AI Technical Summary

Problems solved by technology

This algorithm needs to store the feature database, takes up a lot of storage space, and takes a long time to search and match

When a false target appears in the field of view, it will seriously affect the recognition performance, and this algorithm is extremely sensitive to the position detection error and apparent diameter detection error of the impact crater target, and the robustness is insufficient.

At present, there is no method that can realize the detection and identification of impact craters with large-scale variation range at the same time, and has high accuracy and strong robustness.

Images