A Novel Optimization Method for Lenticular 3-D Display Based on Light Field Decomposion

Three-dimensional lenticular display synthetic image rendering based on light field acquisition

“…When a light ray (green real line) from a point on the t‐axis pass through cylindrical lens, the its direction changes and becomes parallel with the line, which connects the emitting point and C (as the yellow dotted line in Fig. ) . When synthetic image displayed on LCD plane, there are H × W × 3 rays emit from the LCD plane that pass through the lenticular sheet (in Fig.…”

“… II ~ contains information of N points synthetic images. Capture for the light field: Display each II m ~ and photographically acquire images with a moving camera recorded as after capture images' preprocessed (anti‐distortion, 2D projection and interesting region extraction). A nonlinear mapping V sub = M r _ sp ( X ) from radiance value X to acquired image subpixel′ value V sub similarly adopted in the literature is recovered (in Section 2.4) for light field images . Generate the light field: (i) All the light field images in (2) are combined for an emitting light field L ~ ; (ii) The actual camera positions for each movement are exactly calibrated during the photographically acquisition in (2) (in Section 2.2), and the target light field L can be computed according to the actual camera positions. Establish the mapping rules from synthetic image displayed on LCD plane to the viewing plane: We determine s an one‐to‐one mapping (recorded in a look‐up‐table) from points on LCD plane (synthetic image) to viewing plane by rays back‐projection method in the study of Pei et al Compute the optimization matrix: Find the optimization matrix w : min|| L − wL ~ || 2 .…”

“…All the optical paths are calculated by the rays back‐projection method (in Fig. ), and a Look‐Up‐Table LUT 1 records the indexes from points on LCD plane to the points of viewing plane, i.e., …”

“…The key element of lenticular 3D display is to render an accurate synthetic image, which is mapped and synthesized from mulitiview images . A universal rendering rule is to determine the view number of a given point in LCD plane by Philips' method .…”

“…Two different approaches are commonly used to reduce the crosstalk effects: (1) to incorporate extra elements, for example, parallax barriers, pixel masks, or add a timing control scheme; and (2) to correct subpixel values in the synthetic images to minimize the crosstalk. Examples include crosstalk coefficient calculation or weighted value . All of these methods reduce crosstalk in the viewpoints reasonably well.…”