Research On Lenticular-lens Based Multi-view Auto-stereoscopic Display Technology

3D display technology has always been the research hotspot of the field of display technology, because it can provide the audience incomparable 3D vision enjoyment. Previous 3D display technologies need special auxiliary equipment such as glasses and Helmet-Mounted Displays, thus greatly limiting their applications. Multi-view auto-stereoscopic display technology can break through such limitations and give the audience stronger 3D visual impact, which greatly promotes the development and applications of 3D display technology.Compared to other auto-stereoscopic display technologies, the lenticular-lens based multi-view auto-stereoscopic display technology is more practical and has stronger 3D visual impact, which makes it become the research direction and hotspot.This paper studies the lenticular-lens based multi-view auto-stereoscopic display technology. The main innovations of the paper are listed as follows:(1) The paper presents a generalized method of measuring the lenticular parameters, which can be fit for lenticular lens with arbitrary value of LPI (Lines Per Inch) and tilt angle a. This method overcomes the limitations and shortcomings of existing methods.(2) The paper presents a generalized method of synthesizing the 3D image from multi-view images, which can be fit for lenticular lens with arbitrary value of LPI and tilt angle a. This method can synthesize the 3D image with arbitrary size and number of views quickly and efficiently, thus greatly expands the applications of the auto-stereoscopic display technology.(3) The paper presents a generalized method of decomposing the synthesized 3D image into multi-view images. By this method, we can obtain multi-view images with different visual angles from the synthesized 3D image which can be then used for synthesizing the 3D image we need. Therefore, it greatly expands the applications of the 3D image.(4) The paper presents an improved edge adaptive image-scaling algorithm (IEAIS). This algorithm first separates the edges of the image into eight directions according to the absolute value of the first-order difference of the pixels. Then, it uses the improved second-order newton interpolation in horizontal directions, vertical directions and non-edge regions, while the bilinear interpolation in other directions. The results show that the algorithm achieves high-quality images with vivid and sharp edges, and has low computational complexity, which makes it easy for real-time hardware implementation.