Three-dimensional reconstruction method based on multi-view self-adaptation characteristic registration

Abstract

The invention discloses a three-dimensional reconstruction method based on multi-view self-adaptation characteristic registration. The method comprises the following steps that: A, seed points in each two-dimensional picture are selected; B, a region growing method is adopted to perform image segmentation on the two-dimensional picture so as to obtain the boundary point set of each image of each picture; C, through combining the boundary point sets of the segmented images and a polar geometry constraint theory, three-dimensional registration is performed on two images of which the angles are close to each other in a self-adaptation manner; and D, and three-dimensional reconstruction is performed according to the parallax images of boundary points. According to the three-dimensional reconstruction method based on the multi-view self-adaptation characteristic registration of the invention, the region growing method for image segmentation and the polar geometry constraint theory are fully utilized, such that requirements for characteristic point selection and three-dimensional registration can be satisfied. The three-dimensional reconstruction method is applicable to the implementation of three-dimensional reconstruction systems.

Description

Technical field [0001] The invention relates to a computer image processing method, in particular to a method of obtaining three-dimensional image data by processing pixels. Background technique [0002] The key to three-dimensional reconstruction from multi-angle inspection images is to determine the corresponding relationship between the same object point in the scene in different images. One of the methods to solve this problem is to select appropriate image features and match them. Features are pixels or pixel sets or their abstract expressions. Commonly used matching features are mainly point features, linear features, and regional features. Generally speaking, large-scale features contain rich information and have less data and are easy to obtain fast matching, but their extraction and description are relatively complicated, and the positioning accuracy is also poor; while the small-scale features themselves have high positioning accuracy and express description It is simp...

AI Technical Summary

Problems solved by technology

[0003] The present invention mainly aims at the defects of large amount of computation and weak robustness in existing methods for 3D reconstruction of 2D images, and proposes a 3D reconstruction method based on self-adaptive feature registration of multi-view image region growth, It is a process of developing groups of pixels or regions into larger regions by merging adjacent pixels with similar properties as each seed point, such as intensity, grayscale, texture color, etc., into a given area, so it is an iterative process, which can usually segment connected areas with the same characteristics, provide good boundary information and segmentation results, and is easy to implement

Images