Reconstruction and super-resolution algorithms of 2D-3D images and objects

Stereoscopic displays designed to reconstruct 3D objects are still expensive in comparison with conventional ones. So, the possible solution in the 3D perception can be the conversion of 2D to 3D video using optical flow, depth map, etc. extracting needed information that permits the reconstruction of the 3D video sequence [1, 2]. The anaglyph technique is the simplest way to realize 3D perception with less computational coast. To obtain anaglyphs, the left view in blue (or green) color is superimposed on the same image with the right view in red one. When viewed through spectacles of corresponding colors but reversed, the 3D effect can be perceived. Standard techniques employ the optical flow algorithms, matching algorithms, etc. to compute depth maps from a stereo pair or via adjustment of the video frames but they have the drawback: intensive computational operation [3, 4]. The novel approach in the reconstruction of 3D video sequences from 2D ones is consists of following steps: depth map computation, anaglyph construction, anaglyph enhancement, and the 3D video construction from anaglyph obtained. To find the depth map the method of the Minimization of the Global Error Energy [5] has been implemented finding the most reliable estimation of the disparity map via minimization of the error energy as E_d(i,j)=1/3·n·mⁱ⁺ⁿΣ _x=i^j+mΣ_y=j³Σ _k=1(L(x,y+d(i,j)k)-R(x,y,k))², where the left and right images are (L(i, j, k) and R(i, j, k), accordingly. The depth map is formed as follows: DepthMap(i,j)=c√MV_x²(i,j)+MV _y²(i,j), where MV(i, j)_x² and MV(i, j)_y² are the X and Y motion vectors values for each a pixel. Additionally, the dynamic range reduction of the depth map values should be employed to minimize ghosting effects and color losses using the P-th law transformation for dynamic range compression [6, 7]: D_new= a.D ^P.