Research On Efficient Coding Method Based On Depth Offset Mapping For 3D Image Rendering

Nowadays,3D display technology has gained widespread attention from experts as it can show both the plane information and the depth information of the images at the same time.It can further meet people's demands on visual information acquisition.The generation of 3D display content has also become an important part in achieving commercialization of 3D display technology.Due to errors caused by camera parameters and external environment,traditional image acquisition methods cannot fulfill the high demands on 3D information and strict requirements on image accuracy.Thus,it requires additional time consumed on image processing,causing much longer rendering time.It makes real-time rendering difficult to be achieved.The rendering method based on depth image uses the depth information of images to generate virtual viewpoints,which can simplify the process of image acquisition.As the display technology develops rapidly,the resolution of the display devices becomes higher.A lot of information redundancy is caused during the rendering process by using traditional method based on depth image.Furthermore,it also requires a high transmission bandwidth and a much higher demand on the hardware as well,which make it not conducive to achieve system integration and commercialization.Here,an efficient method based on depth offset mapping is proposed in this thesis for 3D image rendering.The applications of this proposed method is verified in various 3D display systems.The main research and achievements of this thesis are as follows,(1)An efficient method based on depth offset mapping for 3D image rendering is proposed.A 2D colored picture is adopted as the reference picture in the proposed method.The corresponding grayscale depth map is used to obtain depth information to calculate the offset of each sub-pixel.Based on the geometric relationship between the display panel and the viewing position,the stereoscopic 3D picture can be rendered directly.As the synthesized 3D picture passes through the display device with respective parameters,the sub-pixels with different offsets can be seen by viewers' eyes with depth perception.Since this method does not require the synthetic virtual viewpoints to be stored in hardware,it reduces the memory requirements of the hardware system.Besides,this method has a fast calculation speed and a high synthesizing efficiency,which can complete the real-time image rendering based on hardware devices with better iterative performance.(2)The implementation of the efficient method based on depth offset mapping in traditional 3D display systems is achieved.The efficient method based on depth offset mapping can be adjusted according to respective structures of various 3D display systems.In this thesis,the application of the proposed method in both flat panel auto-stereoscopic 3D display system and integral imaging 3D display system has been studied.Experiments show that the 3D picture can be rendered by the proposed method in a faster and more efficient way with a good visual perception.An intelligent Android motherboard RK3399 is used in experiments as the hardware device,which is more portable and easier to be industrialized compared to traditional device like PC.It is also more conducive to realize widespread promotion.(3)The implementation of the efficient method based on depth offset mapping in high-dimensional 3D display systems is achieved.In order to enhance the three-dimensional effects in the 3D display system,a high-dimensional 3D display system with double-layered cylindrical lens grating and lens array is adopted.By utilizing the refraction effects of cylindrical lens,it can increase the number of horizontal rays entering human eyes,hence better restoring the light field information and enhancing the three-dimensional impression.The efficient method based on depth offset mapping can be adjusted and applied in high-dimensional 3D display system.Experiments show that the proposed method can achieve good viewing experience with a fast rendering speed.At the same time,the double-layers system has a better three-dimensional experience than the traditional single-layer system does,which further improves the display effect.(4)The implementation of the efficient method based on depth offset mapping in high-resolution 3D display systems is achieved.Screen splicing method is adopted to achieve high-resolution 3D display.In the experiment,four display screens(2×2)are used for splicing,and the viewing zone of each screen has to be adjusted individually so that the viewers can see the correct spliced image from respective positions.The efficient method based on depth offset mapping can be applied on the spliced 3D display system.Multi-screen synchronous display can be performed by adjusting the driving device of each screen.Finally,a spliced high-resolution 3D display system with real-time 3D image rendering is achieved in the experiment.