Image segmentation and adaptive weighting-based stereo matching method

Abstract

The invention discloses an image segmentation and adaptive weighting-based stereo matching method. The method comprises the steps of parallax initialization and parallax optimization. The parallax initialization comprises the steps of adopting corrected left and right images as a reference image and a target image respectively, and segmenting the image; based on the constructed combination segmentation information and an adaptive-weighting cost function, calculating matching costs E (p, pd, d) between a current to-be-matched pixel pint p in the left image and all candidate matching points pd in the right image, and selecting a candidate matching point of a minimum matching cost as an optimal matching point for the pixel pint p; repeating the above steps till all pixel points in the left image are traversed in the raster scan order. In this way, an initial parallax image is obtained. The parallax optimization comprises the steps of fitting the parallax plane of the obtained initial parallax image, suppressing the abnormity of the initial parallax image, and recovering the edge of the initial parallax image. The method is advantaged in that unreliable points in the initial parallax image obtained based on the calculation of the optimal matching point are re-corrected, while abnormal small areas are merged to adjacent normal areas. Meanwhile, edge pixels are recovered, and the parallax error is eliminated. The matching accuracy is improved.

Description

technical field [0001] The invention belongs to the field of image processing, and relates to a stereo matching technology, in particular to a stereo matching method based on image segmentation and adaptive weight. Background technique [0002] In recent years, as one of the hottest research issues in the field of computer vision, stereo vision technology has been widely used in visual navigation, object recognition and industrial control. Stereo vision technology mainly includes image acquisition, camera calibration, feature extraction, stereo matching, and 3D reconstruction. Stereo matching is the core part of stereo vision technology. Correspondence. Whether the images can be accurately matched and the correct three-dimensional coordinates of the scene can be obtained is the key to the success of the stereo vision technology. [0003] According to different constraints, stereo matching can be divided into two categories: global stereo matching and local stereo matching....

AI Technical Summary

Problems solved by technology

For example: a stereo matching method based on confidence support windows, first use the SAD (Sum of Absolute Difference) algorithm to obtain the initial disparity of pixels, and then use pixels with higher confidence in each matching window to perform plane simulation Combined to obtain the final disparity image, although this method can obtain better matching results, it is only suitable for image areas with smooth textures, so there are large limitations; a stereo matching algorithm based on joint histogram, the algorithm The redundant calculation of repeated filtering is reduced by effectively sampling the pixels in the matching window. Although a fixed spatial sampling value can be obtained, the matching result is too sensitive to the characteristics of the input image and the spatial sampling rate. The first method is not universal at present; a stereo matching method based on adaptive window, firstly use the Gaussian mixture model to describe the disparity distribution of the pixels in the matching window, and then determine the size of the matching window according to the uncertainty of the disparity distribution, the method Although the matching quality is improved, it also greatly increases the complexity of the algorithm; a matching method based on adaptive support weights, which does not change the size and shape of the matching window, but chooses a fixed-size rectangular window, according to the window The color and distance difference between each pixel point and the center point are assigned support weights to carry out energy aggregation. This method effectively avoids the problem of matching window selection. Although it can achieve better matching results, it still has the following shortcomings: For the matching window It is difficult to obtain the correct matching result by matching based on pixel color and distance

Images