Research On Stereoscopic Image Editing Based On Image Content

With the hot development of virtual reality(VR)technology,3D imaging technology has been paid attention once again.In addition to being applied to the field of film and entertainment,3D technology is also widely used in these fields,such as industrial design and modeling,simulation and modeling military medical and other related fields.Although there is great process in 3D technology,high quality 3D content is still inadequate,which hinder the development of 3D technology in a large extent.There are some questions due to lack of high 3D content,such uncomfortable 3D viewing,3D content adaptive-display,lack of larger view 3D conten.In this paper,based on binocular 3D image,which is different from the traditional 2D,we analysis the corresponding relationship between the left view and the right view,and expand the conventional image editing techniques to 3D image.We research on a few key questions in 3D image editing.These major contributions of the paper are as follows:First,proposed a Content-aware disparity adjustment method.The disparity mapping is established to control and retarget the depth of a stereoscopic scene.Then,the relationship between the disparity editing and image content editing is established to guide the proposed warping model.At last,to implement the disparity mapping operator,we propose a content-aware stereoscopic mesh warping model,which can simultaneously avoid the salient region distortion and adjust disparity to a target range by establishing the relationship.Second,proposed two content-adaptive stereoscopic image retargeting methods.The first one is a stereoscopic image retargeting method based on Seam Caving.The proposed method defines a new energy function,which include image gradient and image matching factor.The energy function can reduce the distortion of image salient region and keep the stereo consistent.And then we use the dynamic programming method to find the minimum energy seam.This method can reduce image distortion and keep the original disparity.The second one is a retargeting method based on warping method.A rapid estimation method of salient map is proposed by a sparse disparity map.Then,the proposed method via a convex quadratic programming can simultaneously avoid the distortion of the salient region and adjust disparity to a target area,by salient map and energy function.The experimental results show that the proposed method is able to successfully adapt the image disparity to the target display screen,while the salient objects remain undistorted in the retargeted stereoscopic image.Third,proposed a hybrid warping model for stereoscopic image stitching.A uniform homography algorithm is proposed to pre-warp left and right images,and thus ensure consistent changes.And then,a content-preserving warping is introduced to locally refine alignment and reduce vertical disparities.Finally,a Seam-Cutting based algorithm is used to find a blending seam and the multi-band blending algorithm is used to produce the final stitched image.Experimental results show that the proposed method,which not only avoids local distortions,but also reduces vertical disparities reasonably,can effectively stitch stereoscopic images.