ABSTRACT
Tabletop three-dimensional light field display is a kind of compelling display technology that can simultaneously provide stereoscopic vision for multiple viewers surrounding the lateral side of the device. However, if the flat panel light field display device is simply placed horizontally and displayed directly above, the visual frustum will be tilted and the 3D content outside the display panel will be invisible, the large oblique viewing angle will also lead to serious aberrations. In this paper, we demonstrate what we believe to be a new vertical spliced light field cave display system with an extended depth content. A separate optimization of different compound lens array attenuates the aberration from different oblique viewing angles, and a local heating fitting method is implemented to ensure the accuracy of fabrication process. The image coding method and the correction of the multiple viewpoints realize the correct construction of spliced voxels. In the experiment, a high-definition and precisely spliced 3D city terrain scene is demonstrated on the prototype with a correct oblique perspective in 100-degree horizontal viewing range. We envision that our research will provide more inspiration for future immersive large-scale glass-free virtual reality display technologies.
ABSTRACT
The three-dimensional (3D) light field display (LFD) with dense views can provide smooth motion parallax for the human eye. Increasing the number of views will widen the lens pitch, however, resulting in a decrease in view resolution. In this paper, an approach to smooth motion parallax based on optimizing the divergence angle of the light beam (DALB) for 3D LFD with narrow pitch is proposed. DALB is controlled by lens design. A views-fitting optimization algorithm is established based on a mathematical model between DALB and view distribution. Subsequently, the lens is reversely designed based on the optimization results. A co-designed convolutional neural network (CNN) is used to implement the algorithm. The optical experiment shows that a smooth motion parallax 3D image is achievable through the proposed method.
ABSTRACT
Tabletop light field displays are compelling display technologies that offer stereoscopic vision and can present annular viewpoint distributions to multiple viewers around the display device. When employing the lens array to realize the of integral imaging tabletop light field display, there is a critical trade-off between the increase of the angular resolution and the spatial resolution. Moreover, as the viewers are around the device, the central viewing range of the reconstructed 3D images are wasteful. In this paper, we explore what we believe to be a new method for realizing tabletop flat-panel light field displays to improve the efficiency of the pixel utilization and the angular resolution of the tabletop 3D display. A 360-degree directional micro prism array is newly designed to refract the collimated light rays to different viewing positions and form viewpoints, then a uniform 360-degree annular viewpoint distribution can be accurately formed. In the experiment, a micro prism array sample is fabricated to verify the performance of the proposed tabletop flat-panel light field display system. One hundred viewpoints are uniformly distributed in the 360-degree viewing area, providing a full-color, smooth parallax 3D scene.
ABSTRACT
A true-color light-field display system with a large depth-of-field (DOF) is demonstrated. Reducing crosstalk between viewpoints and increasing viewpoint density are the key points to realize light-field display system with large DOF. The aliasing and crosstalk of light beams in the light control unit (LCU) are reduced by adopting collimated backlight and reversely placing the aspheric cylindrical lens array (ACLA). The one-dimensional (1D) light-field encoding of halftone images increases the number of controllable beams within the LCU and improves viewpoint density. The use of 1D light-field encoding leads to a decrease in the color-depth of the light-field display system. The joint modulation for size and arrangement of halftone dots (JMSAHD) is used to increase color-depth. In the experiment, a three-dimensional (3D) model was constructed using halftone images generated by JMSAHD, and a light-field display system with a viewpoint density of 1.45 (i.e. 1.45 viewpoints per degree of view) and a DOF of 50â cm was achieved at a 100 ° viewing angle.
ABSTRACT
Image visual quality is of fundamental importance for three-dimensional (3D) light-field displays. The pixels of a light-field display are enlarged after the imaging of the light-field system, increasing the graininess of the image, which leads to a severe decline in the image edge smoothness as well as image quality. In this paper, a joint optimization method is proposed to minimize the "sawtooth edge" phenomenon of reconstructed images in light-field display systems. In the joint optimization scheme, neural networks are used to simultaneously optimize the point spread functions of the optical components and elemental images, and the optical components are designed based on the results. The simulations and experimental data show that a less grainy 3D image is achievable through the proposed joint edge smoothing method.
ABSTRACT
A holographic function screen (HFS) can recompose the wavefront and re-modulate the light-field distribution from a three-dimensional (3D) light field display (LFD) system. However, the spread function of existing HFSs does not particularly suit integral imaging (II) 3D LFD systems, which causes crosstalk and reduces the sharpness of reconstructed 3D images. An optimized holographic function screen with a flat-top rectangular spread function (FRSF) was designed for an II 3D LFD system. A simulation was carried out through ray tracing, which verified that the proposed diffusion function could suppress crosstalk and improve the overall effect.
ABSTRACT
Three-dimensional (3D) light-field displays (LFDs) suffer from a narrow viewing angle, limited depth range, and low spatial information capacity, which limit their diversified application. Because the number of pixels used to construct 3D spatial information is limited, increasing the viewing angle reduces the viewpoint density, which degrades the 3D performance. A solution based on a holographic functional screen (HFS) and a ladder-compound lenticular lens unit (LC-LLU) is proposed to increase the viewing angle while optimizing the viewpoint utilization. The LC-LLU and HFS are used to create 160 non-uniformly distributed viewpoints with low crosstalk, which increases the viewpoint density in the middle viewing zone and provides clear monocular depth cues. The corresponding coding method is presented as well. The optimized compound lenticular lens array can balance between suppressing aberration and improving displayed quality. The simulations and experiments show that the proposed 3D LFD can present natural 3D images with the right perception and occlusion relationship within a 65° viewing angle.
ABSTRACT
For a floating three-dimensional (3D) display system using a prism type retroreflector, non-retroreflected light and a blurred 3D image source are two key causes of the deterioration in image quality. In the present study, ray tracing is used to analyze the light distribution of a retroreflector at different incident angles. Based on this analysis, a telecentric retroreflector (TCRR) is proposed to suppress non-retroreflected light without sacrificing the viewing angle. A contrast transfer function (CTF) is used to evaluate the optical performance of the TCRR. To improve the 3D image source, the relationship between the root mean square (RMS) of the voxels and the 3D image quality is discussed, and an aspheric lens array is designed to reduce aberrations. Computational simulation results reveal that the structural similarity (SSIM) of the 3D image source increased to 0.9415. An experimental prototype system combining the TCRR and optimized 3D image source is then built. Experimental analysis demonstrates that the proposed method suppresses non-retroreflected light and improves the 3D image source. In particular, a clear floating 3D image with a floating distance of 70 mm and a viewing angle of 50° can be achieved.
ABSTRACT
Lens aberrations degrade the image quality and limit the viewing angle of light-field displays. In the present study, an approach to aberration reduction based on a pre-correction convolutional neural network (CNN) is demonstrated. The pre-correction CNN is employed to transform the elemental image array (EIA) generated by a virtual camera array into a pre-corrected EIA (PEIA). The pre-correction CNN is built and trained based on the aberrations of the lens array. The resulting PEIA, rather than the EIA, is presented on the liquid crystal display. Via the optical transformation of the lens array, higher quality 3D images are obtained. The validity of the proposed method is confirmed through simulations and optical experiments. A 70-degree viewing angle light field display with the improved image quality is demonstrated.
ABSTRACT
Time-multiplexed light-field displays (TMLFDs) can provide natural and realistic three-dimensional (3D) performance with a wide 120° viewing angle, which provides broad potential applications in 3D electronic sand table (EST) technology. However, current TMLFDs suffer from severe crosstalk, which can lead to image aliasing and the distortion of the depth information. In this paper, the mechanisms underlying the emergence of crosstalk in TMLFD systems are identified and analyzed. The results indicate that the specific structure of the slanted lenticular lens array (LLA) and the non-uniformity of the emergent light distribution in the lens elements are the two main factors responsible for the crosstalk. In order to produce clear depth perception and improve the image quality, a novel ladder-type LCD sub-pixel arrangement and a compound lens with three aspheric surfaces are proposed and introduced into a TMLFD to respectively reduce the two types of crosstalk. Crosstalk simulation experiments demonstrate the validity of the proposed methods. Structural similarity (SSIM) simulation experiments and light-field reconstruction experiments also indicate that aliasing is effectively reduced and the depth quality is significantly improved over the entire viewing range. In addition, a tabletop 3D EST based on the proposed TMLFD is presented. The proposed approaches to crosstalk reduction are also compatible with other lenticular lens-based 3D displays.
ABSTRACT
Three-dimensional (3D) light-field display has achieved a great improvement. However, the collection of dense viewpoints in the real 3D scene is still a bottleneck. Virtual views can be generated by unsupervised networks, but the quality of different views is inconsistent because networks are separately trained on each posed view. Here, a virtual view synthesis method for the 3D light-field display based on scene tower blending is presented, which can synthesize high quality virtual views with correct occlusions by blending all tower results, and dense viewpoints on 3D light-field display can be provided with smooth motion parallax. Posed views are combinatorially input into diverse unsupervised CNNs to predict respective input-view towers, and towers of the same viewpoint are fused together. All posed-view towers are blended as a scene color tower and a scene selection tower, so that 3D scene distributions at different depth planes can be accurately estimated. Blended scene towers are soft-projected to synthesize virtual views with correct occlusions. A denoising network is used to improve the image quality of final synthetic views. Experimental results demonstrate the validity of the proposed method, which shows outstanding performances under various disparities. PSNR of the virtual views are about 30â dB and SSIM is above 0.91. We believe that our view synthesis method will be helpful for future applications of the 3D light-field display.
ABSTRACT
Three-dimensional (3D) light field displays require samples of image data captured from a large number of regularly spaced camera images to produce a 3D image. Generally, it is inefficient to generate these images sequentially because a large number of rendering operations are repeated in different viewpoints. The current 3D image generation algorithm with traditional single viewpoint computer graphics techniques is not sufficiently well suited to the task of generating images for the light field displays. A highly parallel multi-view polygon rasterization (PMR) algorithm for 3D multi-view image generation is presented. Based on the coherence of the triangular rasterization calculation among different viewpoints, the related rasterization algorithms including primitive setup, plane function, and barycentric coordinate interpolation in the screen space are derived. To verify the proposed algorithm, a hierarchical soft rendering pipeline with GPU is designed and implemented. Several groups of images of 3D objects are used to verify the performance of the PMR method, and the correct 3D light field image can be achieved in real time.
ABSTRACT
A novel optical reverse mapping (ORM) method and an ORM criterion are proposed to evaluate the relevance between the directional backlight (DB) 3D light-field display system aberration and the crosstalk. Based on the ORM criterion, the space-division-multiplexed catadioptric integrated backlight (SCIB) and symmetrical triplet-compound lenticular array (triplet LA) are designed. The SCIB is composed of hybrid Fresnel integrated backlight unit (hybrid Fresnel unit) and space-division-multiplexed microprism unit (microprism unit). The hybrid Fresnel unit is used to provide the directional light, and the divergence angle is 2.4-degrees. The average uniformity of 83.02% is achieved. The microprism unit is used to modulate the directional light distribution into three predetermined directions to establish a 90-degree viewing area. Combined with SCIB, the triplet LA is used to suppress the aberrations and reduce the crosstalk. In the experiment, a DB 3D light-field display system based on SCIB and triplet LA is set up. The displayed light-field 3D image can be observed in a 90-degree viewing angle. Compared to the conventional DB 3D display system, the light-field 3D image is aberration-suppressed, and the SSIM values are improved from 0.8462 to 0.9618. Meanwhile, the crosstalk measurement results show that the average crosstalk is 3.49%. The minimum crosstalk is 2.31% and the maximum crosstalk is 4.52%. The crosstalk values in 90-degree are lower than 5%.
ABSTRACT
A novel catadioptric retroreflector floating device (CRA) used in the 3D floating light-field system is proposed. The floating light-field image constructed by the CRA is aberration-suppressed. The luminance and the contrast of the image are substantially improved in a 90-degree viewing angle. The CRA is constituted of the designed catadioptric retroreflector (CR). The CR consists of three lenses, the first and the second lens is to refract the light, and the rear surface of the third lens is coated with reflective coating in order to reflect the incident light. The CRA is processable and the fabrication process using UV embossing is also described. A spectrophotometer is utilized to measure the retroreflective efficiency of the CRA. The average retroreflective efficiency of the CRA is 80.1%. A beam quality analyzer is utilized to measure the beam spot quality of the CRA, and the image quality can satisfy the requirements of human eye observation. In the experiment, compared to the floating light-field image constructed by the micro-beads type retroreflector floating device (MRA), the image quality of the floating light-field image constructed by the CRA is significantly enhanced. In the quantitative computer simulation, the PSNR values of the images are increased from 23.0185 to 32.1958.
ABSTRACT
Due to lack of the accommodation stimulus, an inherent drawback for the conventional glasses-free stereoscopic display is that precise depth cues for the human monocular vision is rent, which results in the well-known convergence-accommodation conflict for the human visual system. Here, a super multi-view light field display with the vertically-collimated programmable directional backlight (VC-PDB) and the light control module (LCM) is demonstrated. The VC-PDB and the LCM are used to form the super multi-view light field display with low crosstalk, which can provide precisely detectable accommodation depth for human monocular vision. Meanwhile, the VC-PDB cooperates with the refreshable liquid-crystal display panel to provide the convergence depth matching the accommodation depth. In addition, the proposed method of light field pick-up and reconstruction is implemented to ensure the perceived three dimensional (3D) images with accurate depth cues and correct geometric occlusion, and the eye tracker is used to enlarge the viewing angle of 3D images with smooth motion parallax. In the experiments, the reconstructed high quality fatigue-free 3D images can be perceived with the clear focus depth of 13â cm in the viewing angle of ± 20°, where 352 viewpoints with the viewpoint density of 1â mm-1 and the crosstalk of less than 6% are presented.
ABSTRACT
The light field display can provide vivid and natural 3D performance, which can find many applications, such as relics research and exhibition. However, current light field displays are constrained by the viewing angle, which cannot meet the expectations. With three groups directional backlights and a fast-switching LCD panel, a time-multiplexed light field display with a 120-degree wide viewing angle is demonstrated. Up to 192 views are constructed within the viewing range to ensure the right geometric occlusion and smooth parallax motion. Clear 3D images can be perceived at the entire range of viewing angle. Additionally, the designed holographic functional screen is used to recompose the light distribution and the compound aspheric lens array is optimized to balance the aberrations and improve the 3D display quality. Experimental results verify that the proposed light field display has the capability to present realistic 3D images of historical relics in 120-degree wide viewing angle.
ABSTRACT
Integral imaging is a promising 3D visualization technique for reconstructing 3D medical scenes to enhance medical analysis and diagnosis. However, the use of lens arrays inevitably introduces flipped images beyond the field of view, which cannot reproduce the correct parallax relation. To avoid the flipping effect in optical reconstruction, a twice-imaging lens array based integral display is presented. The proposed lens arrangement, which consists of a light-controlling lens array, a field lens array and an imaging lens array, allows the light rays from each elemental image only pass through its corresponding lens unit. The lens arrangement is optimized with geometrical optics method, and the proposed display system is experimentally demonstrated. A full-parallax 3D medical scene showing continuous viewpoint information without flipping is reconstructed in 45° field of view.
ABSTRACT
When employing the light field method with standard lens array and the holographic functional screen (HFS) to realize the tabletop three-dimensional (3D) display, the viewing area of the reconstructed 3D images is right above the screen. As the observers sit around the table, the generated viewpoints in the middle of the viewing area are wasteful. Here, a 360-degree viewable light-field display system is demonstrated, which can present 3D images to multiple viewers in ring-shaped viewing range. The proposed display system consists of the HFS, the aspheric conical lens array, a 27-inch LCD with the resolution of 3840×2160, the LEDs array and the Fresnel lens array. By designing the aspheric conical lens, the light rays emitting from the elemental images forms the viewpoints in a ring-type arrangement. And the corresponding coding method is given. Compared with the light field display with standard lens array, the viewpoint density is increased and the aliasing phenomenon is reduced. In the experiment, the tabletop light-field display based on aspheric conical lens array can present high quality 360-degree viewable 3D image with the right perception and occlusion.
ABSTRACT
Three-dimensional (3D) light field display, as a potential future display method, has attracted considerable attention. However, there still exist certain issues to be addressed, especially the capture of dense views in real 3D scenes. Using sparse cameras associated with view synthesis algorithm has become a practical method. Supervised convolutional neural network (CNN) is used to synthesize virtual views. However, such a large amount of training target views is sometimes difficult to be obtained and the training position is relatively fixed. Novel views can also be synthesized by unsupervised network MPVN, but the method has strict requirements on capturing multiple uniform horizontal viewpoints, which is not suitable in practice. Here, a method of dense-view synthesis based on unsupervised learning is presented, which can synthesize arbitrary virtual views with multiple free-posed views captured in the real 3D scene based on unsupervised learning. Multiple posed views are reprojected to the target position and input into the neural network. The network outputs a color tower and a selection tower indicating the scene distribution along the depth direction. A single image is yielded by the weighted summation of two towers. The proposed network is end-to-end trained based on unsupervised learning by minimizing errors during reconstructions of posed views. A virtual view can be predicted in a high quality by reprojecting posed views to the desired position. Additionally, a sequence of dense virtual views can be generated for 3D light-field display by repeated predictions. Experimental results demonstrate the validity of our proposed network. PSNR of synthesized views are around 30dB and SSIM are over 0.90. Since multiple cameras are supported to be placed in free-posed positions, there are not strict physical requirements and the proposed method can be flexibly used for the real scene capture. We believe this approach will contribute to the wide applications of 3D light-field display in the future.
ABSTRACT
A method for the viewing angle and viewing resolution enhancement of integral imaging (InIm) based on time-multiplexed lens stitching is demonstrated using the directional time-sequential backlight (DTS-BL) and the compound lens-array. In order to increase the lens-pitch of the compound lens-array for enlarging the viewing angle of InIm, DTS-BL is used to continuously stitch the adjacent elemental lenses in the time-multiplexed way. Through the compound lens-array with two pieces of lens in each lens unit, the parallel light beams from the DTS-BL converge and form a uniformly distributed dense point light source array (PLSA). Light rays emitting from the PLSA are modulated by the liquid crystal display (LCD) panel and then integrated as volumetric pixels of the reconstructed three-dimensional (3D) image. Meanwhile, time-multiplexed generation of the point light sources (PLSs) in the array is realized by time-multiplexed lens stitching implemented with the DTS-BL. As a result, the number of the PLSs, as the pixels of the perceived 3D image, is increased and then the viewing resolution of the 3D image is obviously enhanced. Additionally, joint optical optimization for the DTS-BL and the compound lens-array is used for suppressing the aberrations, and the imaging distortion can be decreased to 0.23% from 5.80%. In the experiment, a floating full-parallax 3D light-field image can be perceived with 4 times the viewing resolution enhancement in the viewing angle of 50°, where 7056 viewpoints are presented.
