A panoramic image mosaics algorithm based on wavelet decomposition and equidistant matching

A panoramic image mosaics algorithm based on wavelet decomposition and equidistant matching is proposed in this paper. Firstly, the images are decomposed with wavelet, and then the similarity distance between the corresponding pixels is defined. Secondly, according to the criterion of minimizing the sum of similarity distance, the equidistant matching is used at the every level of the wavelet pyramid to align all the images. Finally, all the aligned images are projected onto a plane or cylinder to construct a panoramic image. To further improve the image mosaics quality, a fade in-out blending approach is also employed in this paper. The algorithm is fairly stable and efficient for the use of the edges values and pyramid structure of wavelet decomposition respectively.