Stereo matching algorithm for motion blur and illumination change image

Abstract

The present invention provides an accurate and practical stereo matching algorithm for the problems of motion blur and illumination change in practical acquisition of image, mainly comprising the following parts: 1. providing a regular difference image feature with illumination invariant to solve illumination change between a left image and a right image; 2. establishing a maximum posterior probability formula based on the character that the value of channel alpha changes continuously, modeling and estimating the influence of blur part to the sum of parallax; and 3. providing a strategy basedon symmetrical stereo matching to calculate the accurate value of parallax.

Description

technical field [0001] The invention relates to the technical fields of computer vision and image processing, in particular to stereo matching of motion blur and illumination changes of left and right viewing angles in actual image acquisition. Background technique [0002] Stereo vision has been extensively studied since the emergence of the field of computer vision, because it has many important applications, such as 3D scene reconstruction, machine navigation, 3D television, etc. Stereo vision is to obtain the left and right perspective images of the scene through two cameras (stereo cameras) that are relatively close to each other, and use the related methods of epipolar geometry to achieve horizontal alignment, that is, the corresponding points of the left and right views are on the same horizontal line, and then Accurate coordinate relationship of corresponding points is obtained through matching algorithm. [0003] The research on stereo matching has never been inter...

AI Technical Summary

Problems solved by technology

[0004] However, most of the current stereo matching algorithms focus on processing some public test data sets, and these data are collected under relatively ideal conditions, so the influence of some actual shooting conditions is ignored, among which motion blur and Left and right perspective light changes

Using the existing stereo matching algorithm to process images with motion blur and illumination changes can not get very ideal results

On the one hand, motion blur increases the area of ​​the motion area, and unreasonable disparity values ​​will appear during the matching process. Generally, it is necessary to perform motion recovery first and then perform matching; on the other hand, the current algorithms assume that the left and right view images are in the Acquired under the same lighting conditions, so the effect on lighting changes is very poor

Images