Contour key point registration and identification method based on stable area

Abstract

The present invention discloses a contour key point registration and identification method based on a stable area, and belongs to the technical field of image automatic detection. The method comprises: by adopting the maximum stable area detection method based on histogram statistics, detecting an outer contour of a body image; by adopting an outer contour geometric invariant significant point detection and area registration method, detecting and registering each limb area; and by adopting a area contour registration parameter estimation and location mapping method, implementing localization and segmentation of each tissue and organ area in a human body. According to the method provided by the present invention, in combination with the histogram statistics-based adaptive threshold segmentation, maximum stable area detection, area contour registration and organ location mapping methods, the complexity of maximum stable area detection is greatly reduced, the stability of an image matching algorithm is strengthened, and the accuracy of body organ localization is improved. As shown by results of tests in different image sets, body organ registration and localization by the method provided by the present invention are stabler and more accurate compared with those in an existing method.

Description

technical field [0001] The invention belongs to the technical field of automatic image detection, and more specifically relates to a method for registering and identifying contour key points based on a stable region. Background technique [0002] Image registration is a very important research topic in target detection and recognition, and it is widely used in medical imaging disease detection, fault image detection and matching navigation and other fields. In medical evaluation images with different imaging mechanisms such as B-ultrasound, CT, MRI, PET, SPECT, and IR, in addition to being affected by changing factors such as scale, viewpoint, environment, and shape nonlinear deformation, body tissues also attenuate different electromagnetic signals. There are also great differences in photoelectric sensor characteristics such as radiation, conduction, diffusion, and density, and there are significant differences between different observation images of the same person, which...

AI Technical Summary

Problems solved by technology

[0005] However, when using stable regions and boundaries for image registration, there are several problems that need to be solved: (1) The number and area of ​​tissues in the body are large, and the time complexity of stable region extraction and calculation is high; (2) ) The non-rigid connection of each limb area of ​​the human body leads to nonlinear geometric distortion among tissues in the body; (3) There are great differences in the tissue texture distribution characteristics of different imaging mechanisms, resulting in inaccurate positioning of body tissues

Images