Remote sensing image registration based on contextual local perception and feature fusion

In order to solve the problems of complex and changeable remote sensing image information, low accuracy of registration model, and long time consumption, this paper proposes a remote sensing image registration model based on context local perception module and adaptive channel feature fusion. This method uses context local perception Convolution modules, with a strategy of sharing context-aware weights, thereby capturing finer and richer local features. At the same time, an adaptive feature fusion attention mechanism is introduced to fuse multi-layer features to obtain higher registration accuracy. First, in the feature extraction stage, feature information is extracted from the image through the improved CloFormer model. Secondly, in the matching stage, the cross-correlation two-way matching method is used to calculate the dense correspondence to obtain the two-way parameters, and the final transformation parameters are weighted and synthesized through the parameter regression network, and finally the affine transformation completes the image registration. Experimental results prove that this model effectively improves the accuracy and efficiency of remote sensing image registration.