Image Rectification Schemes Of Multi-view Auto-stereoscopic Video Capture System

Free viewpoint television as the most important multi-view imaging applicationsprovide users the ability to interactively select scene viewing angle, breaking therestrictions which can only be viewed from the camera viewpoint, and enhance theuser experience provided by traditional media. Now it is more and more widespreadconcerned. This thesis ifrst introduces the development of multi-viewauto-stereoscopic television technology at home and abroad, and then proposed aminimum distortion rectiifcation algorithm as the pre-processing method forzero-parallax rectiifcation, a sparse disparity range is generated by blocksegmentation, block matching algorithm and sparse matching algorithm, and ifnallyremaps the original parallax range determined by the parallel cameras array to aproper parallax range according to the research of3D viewing comfort, which greatlyimprove the auto-stereoscopic viewing experience, major achievements are asfollows:(1)The thesis presents a new rectiifcation method for calibrated multi-viewimages based on parallel conifguration mode to reduce geometric distortion. Theoriginal images are re-projected to ideal images which fulifll the conditions withoutvertical parallax by devising an algorithm according to the intrinsic and externalparameters of camera.(2)The thesis also proposes a new rectiifcation method for zero parallax to takeadvantage of the parallax distribution characteristics of human eyes，cameras andauto-stereoscopic displays based on recent researches on the areas of3D viewingcomfort to improve the3D viewing experience. All the algorithms are devised to meetthe needs of real-time implement of multi-view video capture system. They are thecompromise between the precision of calculation and the speed of execution.Experimental results show that the method can implement image rectiifcation andzero parallax rectiifcation in real-time for the multi-view images and support eightviewpoints with naked eyes, and avoid the errors of vertical parallax and improperzero parallax, and improve the effect of auto-stereoscopic displaying.