A point feature-based cylindrical image mosaic method
Xi Li, Ping Gong · 2011
A cylindrical image mosaic method based on Harris corner feature is proposed in the paper. Firstly, the sequence of photographs taken around 360 degree is projected to cylinder coordinate system, and then the Harris corners are detected by an improved Harris corner detecting algorithm. After the NCC (Normalized Cross Correlation) corner matching algorithm performed a rough match, an improved RANSAC (Random Sample Consensus) algorithm is applied twice to remove the mismatched points as far as possible to improve the correct matching ratio. Finally, the image mosaic and fusion are accomplished according to the registration result. Experiments demonstrate that the proposed algorithm has achieved a well result.