Differential Evolution-Based Sample Consensus Algorithm for the Matching of Remote Sensing Optical Images With Affine Geometric Differences

Optical image matching has been a recent trend in the field of remote sensing image processing. It is considered as a challenging problem due to the existence of significant geometric variations as well as intensity differences between the images. Scale invariant feature transform (SIFT) is one of the most effective schemes to handle these factors. However, it produces many false matches in the matching of the remote sensing images which effect its performance. In order to address this issue, a novel Differential Evolution-based Sample Consensus Algorithm (DESCA) is proposed to eliminate these false matches and retain the correct matches. The proposed DESCA scheme is very effective for the images having significant affine geometric differences. It has the ability to provide more correct matches. Several sets of remote sensing optical image pairs are used to test the performance of the proposed method. It obtains the Root Mean Square Error (RMSE) value in the range of 0.67 to 0.95 pixels which indicates that the sub-pixel accuracy is achieved. The experimental results show that the proposed method provides more correct matching pairs and better mutual information (MI) values than the state-of-the-art methods.