An Improved Image Registration Method Using E-SIFT Feature Descriptor with Hybrid Optimization Algorithm

Automatic image registration of the satellite images aids in the field of the computer-aided research. In the recent years, image registration is useful in the environment monitoring and the agriculture purpose. In this work, the automatic image registration model has been developed through the novel hybrid optimization algorithm. This work mainly concentrates in image registration of the hyperspectral images arriving from the satellite. The proposed automatic image registration model uses the input image and the reference image for the registration purpose. From both the images, the E-SIFT features are extracted and given to the point matching algorithm for the keypoint detection. Then, the similarity transformation model gets the keypoints and makes the input images to the original position. Here, the weighted average model is developed for the image fusion, and the weight score for the image fusion is selected optimally through the proposed salp swarm-crow search algorithm (SS-CSA). For the experimentation, the proposed scheme uses the standard database having the hyperspectral satellite images. The simulation results reveal that the proposed image registration scheme with the SS-CSA algorithm has progressed better than the existing techniques with 0.711788 and 0.993602 for the RMSE and NCC, respectively.