Research On Method For Converting2D To3D Video

With the development of digital video technology, digital video is gradually formtwo-dimensional transition to three-dimensional. At present, the insufficient3D sourceis the bottleneck of the development of3D video. Stereoscopic film requires specifictechnology from shooting to issue, and it need high costs and long time. How to convertthe existing2D video into3D video has become an important research field in the3Dtechnology.In this paper, the format of "two-dimensional video+depth map" is used toconvert2D into3D video. A depth map for the video frame is mainly studied. Gaussianbackground modeling method is used to extract the foreground moving objects. A depthvalue allocation method is proposed based on the assumption that the gradient depth hasnothing to do with the scene: based on grey value128, gray value of0to255can bedivided into two ranges which are from0to127and from128to255. According to thegradient value distribution formula, the gradient depth of background region is assigned.Then, according to the motion information of moving objects in video, the depth offoreground area is assigned. Finally, a depth map of the video frame is generated.For the disparity estimation, the methods of disparity estimation based on theregional and based on the characteristics are analyzed. The depth value128is assignedto the zero disparity based on the depth map. The depth value that is more than128isassigned to negative parallax value, and the depth value that is less than128is assignedto positive parallax value. Thus, transformation the depth map into the disparity map iscompleted.For the problem of the left image with holes, a method of depth image basedrendering for red and blue3D image construction is proposed. The red primary image isshifted left to obtain left image based on the parallax information. Because the leftimage has holes, the depth map smoothing and improved hole filling technology areused to fill the holes effectively and the image quality is improved, which reducescomputational complexity of hole filling. An effective method for converting2D to3Dis acquired.