Registration Method of Image Set Based on Dynamic Directed Graph

Abstract

The invention discloses a method for registering an image set based on dynamic digraphs. The method includes the steps that (1) a self-adaptation weighing sparse representation method is utilized to calculate global similarity of images in the image set and establish or update the digraphs with the images in the image set being nodes; (2) the shortest path between every two nodes and root node images are determined based on the digraph result, and each node image is registered to the direct father node image of each node image according to the shortest path from each node image to the corresponding root node image, wherein the shortest path is the registration path; (3) a weight table used in the self-adaptation weighing sparse representation method is updated; (4) the first step, the second step and the third step are repeated until all the images are registered to a same root node image. The self-adaptation weighing sparse representation method is utilized to calculate the global similarity of the images, and the dynamic digraphs are combined, so that the image set is accurately and efficiently registered.

Description

technical field [0001] The invention relates to the field of image processing, in particular to an image set registration method based on a dynamic directed graph. Background technique [0002] Image set registration (Group-wise image registration) plays a very important role in medicine, such as the detection of brain region function, monitoring of lesion development (registration of lesion images in different periods to obtain the development trend of the lesion), and surgery. Monitoring of postoperative organ recovery (registration of images of postoperative organs obtained at different periods to obtain healing progress). At present, the methods in the field of image set registration can be roughly divided into two categories: the first category is the registration method directly based on pairwise images, and the second category is the registration method based on the intermediate template. [0003] Most of the image set registration methods proposed earlier are direct...

AI Technical Summary

Problems solved by technology

However, due to the registration error between the two images, the use of formula (2) will cause the registration error to accumulate

In addition, the measurement of image similarity is still based on calculating the similarity between two images, ignoring the global image information, resulting in the generated image not being able to estimate the manifold of the image set well, making the registration path calculated later unsatisfactory.

Images