Robust Image Matching under a Large Disparity

We present a new method for detecting point matches between two images that have a large disparity resulting from camera rotations and zooming changes. Our strategy is to impose various constraints such as local image correlations, spatial consistency, and global smoothness as “soft” constraints via their “confidence” before imposing the “hard” epipolar constraint by RANSAC. We also introduce a model selection procedure to test if the image mapping can be regarded as a homography. We demonstrate the effectiveness of our method by real image examples.