Complex parallax stereo matching method, device and equipment based on forward and backward smoothing and O (1)

Abstract

The invention discloses a complexity parallax stereo matching method based on forward and backward smoothing and O (1), which comprises the following steps of respectively carrying out forward and backward smoothing processing on left and right eye images, constructing a cost calculation function by combining color and gradient information, and calculating the cost values, at the cost aggregationphase, respectively constructing a minimum spanning tree for the smoothed left-eye image and the smoothed right-eye image; performing cost aggregation on the cost function value; obtaining an initialparallax error by adopting a WTA strategy; judging stable and unstable points through left-right consistency detection, obtaining initial parallax confidence, carrying out hole filling on the unstablepoints to obtain an initial parallax image, combining color information of the left image with the initial parallax image to obtain a mixed weight, and carrying out confidence aggregation on the confidence value by adopting a horizontal tree structure based on the initial parallax confidence and the mixed weight; carrying out belief propagation on the belief aggregation value to obtain an optimaldisparity estimation value, thereby obtaining a final dense disparity map. According to the invention, the accuracy and the efficiency of stereo matching are effectively improved.

Description

technical field [0001] The invention belongs to the technical field of image processing, and relates to a stereoscopic matching method based on forward-backward smoothing and O(1) complexity parallax. Background technique [0002] Stereo matching algorithms are widely used in computer vision, such as 3D reconstruction, image focusing, etc., but there are still many challenging problems. The main work of stereo matching is to find the corresponding image point pairs in the image, including four steps: matching cost calculation, cost aggregation, disparity calculation and disparity refinement. Algorithms are generally divided into global and local algorithms. [0003] The purpose of the global algorithm is to minimize the energy function of the matching problem. The energy function includes data items and smoothing items. When the disparity values ​​of adjacent nodes are very different, the smoothing item is used as a penalty factor. Global algorithms mainly include methods ...

AI Technical Summary

Problems solved by technology

In the process of parallax refinement based on MST, the cost value of the unstable point is updated by propagating the value of the stable point to the unstable point. The complexity of any pixel is also O(d), and the complexity is relatively high.

In addition, a small amount of noise in the image will affect the matching accuracy

Images