Non-rigid registration method and system for maximum moment and space consistency of multimode image

Abstract

The invention discloses a multimode image maximum moment and space consistency non-rigid registration method and system, and the method comprises the steps: constructing a global transformation sub-network and a deformation attention sub-network, and constructing a multimode image non-rigid registration network capable of end-to-end training in combination with a position transformation grid and a pixel resampling layer; constructing a loss function for the multimode image non-rigid registration network; and constructing a training data set by using the multimode image, and training the multimode image non-rigid registration network by using the constructed training data set and the loss function. According to the method, the distorted image without geometric correction can be directly registered, the problem of local distortion of the multimode image is well solved, accurate registration of the multimode image is realized, reliable support can be provided for accurate image fusion and accurate target detection, and the method can be applied to the application fields of natural disaster monitoring, resource investigation and exploration, accurate target strike and the like, and has wide application prospects. The system has the advantages of intelligent manufacturing, rescue and relief work, remote sensing monitoring and the like, and is wide in application range.

Description

technical field [0001] The invention belongs to the image registration technology in the field of image processing, and in particular relates to a non-rigid registration method and system for the maximum moment and spatial consistency of a multimodal image. Background technique [0002] Image registration refers to the process of transforming two or more images obtained at different times and under different conditions (weather, illumination, camera angle and position, etc.) from different coordinate systems into the same coordinate system. With the progress of modern science and technology, the development requirements of accuracy and refinement are becoming more and more urgent, and the single-modal image cannot fully express the target. Therefore, the advantages of aggregating multi-modal images can achieve the accurate representation of the target. In the civilian field, Such as visual navigation of intelligent manufacturing, geological disaster monitoring and treatment,...

AI Technical Summary

Problems solved by technology

However, limited by the shooting conditions, terrain environment and camera hardware performance, the problems of large radiation differences, geometric distortion, multi-field of view, multi-scale, multi-temporal, and less data pose challenges to the current image registration algorithm based on deep learning. Serious challenges

Images