Font Size: a A A

The Application Research Of Binocular Vision In Automatic Reversing System

Posted on:2010-06-07Degree:MasterType:Thesis
Country:ChinaCandidate:J WeiFull Text:PDF
GTID:2178360275978659Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
With the sustained Chinese economic growth and living standard improvement , more and more families hold their own cars. Highways, streets, parking lots and garages become overcrowded, more and more hidden dangers brought about by reversing made people headaches, people are looking for automatic reverse.From many sources of information, improvement of automatic reverse requires proper visual capture of information in the environment. How to get the extract accurate three-dimensional is a very important research topic. Based on analyzing the key technologies of automatic reverse, we researched the usage of binocular vision in the automatic reverse system and constructed the experimental system of binocular stereo vision. The research experiments focused on four aspects namely: camera calibration, feature extraction, stereo matching and distance calculating.In this paper, completed the following major tasks:Based on stereo camera, discussed a number of commonly used coordinate system and the conversion for each coordinate system. Gave the model about optical axis parallel binocular vision camera, described a step-by-step calibration of the binocular stereo vision based on coplanar circle in detail with experimental results.According to the easy image texture, point feature was taken as matching primitives. First, the paper described SUSAN and Harris corner detection, by the experimental comparison, selected base on the Harris Sub-pixel Corner detection algorithm for image feature extraction.Took corner points of target geometry as matching primitives, proposed the feature points matching combined with regional correlation matching as stereo matching method. First of all, rough matching, then combined with the epipolar constraint of corresponding points relaxed iterative algorithm was applied, false matching points were removed, until one-to-one matching points were achieved. On this basis, we used the parallax of corresponding points to calculate the target depth. A summary of the application of binocular vision in automatic reversing system was also presented.
Keywords/Search Tags:camera calibration, corner detection, three-dimensional matching, parallax
PDF Full Text Request
Related items