A Method of Graph Distortion Analysis Based on Invariant Moments

Abstract

A method for analyzing image distortion based on invariant moments, comprising: 1) selecting an appropriate moment invariant: in a dynamic body model, using a point distribution model to represent the object outline in an image; 2) an active detection method for model distortion: 2.1 ) For the PDM in the dynamic shape model, each label point has a corresponding weight, and a PDM is decomposed into sub-graphs according to the weight; 2.2) Calculate each sub-graph and model in the training set after each The boundary moment invariants of a sub-graph; then calculate the cluster center values ​​of these boundary moment invariants; 2.3) The steps of carrying out the distortion detection of the model are: 1. divide the deformed model into blocks according to the pre-defined block strategy; ② Calculate the boundary moment invariant of each sub-graph; ③ Calculate the distortion probability of each sub-graph according to formula (2)(3)(4); 3) Active correction method of model distortion. The invention can effectively detect and correct the distortion in the search process, and has good reliability.

Description

technical field [0001] The present invention relates to the field of graphics and images, especially a method for analyzing graphics distortion. The method combines dynamic body model methods, image processing, computer graphics, medicine, mathematics, applied physics, statistics, biology and other technologies, and is mainly used for Analyze image fitting quality, especially for shape segmentation and morphological distortion analysis in medical images. Background technique [0002] Deformable model is an effective image segmentation, registration and recognition technology developed in the late 1980s. Deformable models have the ability to change shapes, which can represent flexible objects. Applying the deformable model to the image can find objects to be segmented, registered or recognized on the image through deformation. Deformable model methods generally consist of two phases, the training phase and the actual matching phase. [0003] Generally, there are two ways t...

AI Technical Summary

Problems solved by technology

[0009] In order to overcome the shortcomings of existing dynamic shape model methods that cannot detect and correct distortion and have poor reliability during the search process, the present invention provides a method based on invariant moments that can effectively detect and correct distortion during the search process and has good reliability. Graphic Distortion Analysis Method

Images