| 摘要 |
Methods for the registration of images typically assume that there are only two images, an assumption that is not always valid. By using the remaining images to obtain a choice of paths between two selected images, the transformation between the two can be determined with greater accuracy by averaging those paths. When averaging the paths, greater weight can be given to paths whose accuracy is known or reasonably believed to be greatest. Iteration of the process may be possible, where the available computation power is available. |
