Study On Technology Of Visual Comfort Enhancement For Stereoscopic Viewing Based On Improved Vertical Parallax Correction

With the improvement of living standards,people are no longer content to obtain the single image from black and white TV but moreover wish to get three-dimensional image by digital TV.Therefore video development underwent the process changing from two-dimensional image to three-dimensional image,the stereo display technology emerged as the times require.For the reason that the three-dimensional video technology is not mature,although people obtained much richer information and more intuitive feelings from stereoscopic video digital TV,but investigation reveals that most watchers have the reaction of visual fatigue such as dizziness,sick and nausea.Therefore,how to conquer the fatigue problem of dizziness,diplopia and blurred vision by studying on anti-fatigue technology became a hot spot in today's research field.A large amount of experimental data proves there are three aspects to arouse visual fatigue: external factors,physiological factors and psychological factors.The physiological factors include the inconsistency between convergence and focus adjustment,vertical disparity and motion parallax.To achieve the goal that the collected images have no vertical parallax,proper correction are taken for these images.The correction methods are divided into two: with camera calibration and without camera calibration.Comparing to the correction method with camera calibration,the latter has much wider applications.However,there are few domestic researches on this method.In this thesis,the vertical parallax reduction methods which can reduce visual fatigue are studied.Under the conditions of unknown camera internal and external parameters,the improvement of the vertical parallax reduction method is achieved without affecting the horizontal parallax.First of all,the significance and the research status quo of stereo visual comfort are briefly introduced.Secondly,a number of factors which contribute to visual fatigue are described and the basic theories of the study are introduced,including the principle of binocular stereoscopic vision system,the nonlinear solution to the uniformity matrix and adaptive interpolation algorithm.Finally,the stereo view comfort enhancement methods based on vertical parallax reduction are proposed:(1)Method of vertical parallax reduction combined with Levenberg-Marquardt algorithm: Firstly,the SIFT algorithm,which is invariant to scaling,rotation and affine transformation,is used to detect the feature matching points from the binocular images.Then according to the coordinate position of matching points,the transformation matrix,which can reduce the vertical parallax,is calculated using Levenberg-Marquardt algorithm.Finally,the transformation matrix is applied to target image to calculate the new coordinate position of each pixel from the view images.(2)Method of vertical parallax reduction combined with Sobel adaptive interpolation: the self-adaptive scaling algorithm is used for the original image interpolation.When hen the pixel point of original image is in the edge areas,the interpretation is implemented adaptively along the edge direction obtained by Sobel operator;otherwise,when the pixel point of original image is not in the edge areas,bicubic interpolation is chosen to interpolate.Then,the interpolation image is processed using the method of vertical parallax reduction combined with Levenberg-Marquardt algorithm.Both improved method and original method are simulated using multiple sets of multi-view video sequences.The experimental results show that compared with the method that can reduce the vertical parallax using linear algorithm to calculate two-dimensional projective transformation,the improved method can obtain better effect in the same aspects of vertical parallax reduction and horizontal parallax correction as well,and method of vertical parallax reduction combined with Sobel adaptive interpolation can obtain better effect than that of applying the method of vertical parallax reduction with combined algorithm of Levenberg-Marquardt.Furthermore the greater the degree of image rotation,the more obvious and evident for the adjustment result through the vertical parallax reduction based on Sobel adaptive interpolation.