Research of SIFT Algorithm Based on Underground Environment

The paper put forward a new feature matching algorithm to solve the problem that existing feature matching methods are not suitable to underground environment. Firstly, the algorithm uses curvelet algorithm to preprocess image noise, then adopts SIFT algorithm to form and match feature vectors, which removes mismatched points effectively by optimization of RANSAC method and achieves image stitching by building projection transformation model. The experiment showed that the algorithm has better robustness in noise, poor illumination and fuzzy environment, and solves mismatching phenomenon in mine environment, whose stitching result is smoothing and natural.