Viewpoint-Predicting-Based Remote Rendering on Mobile Devices Using Multiple Depth Images

In order to break through the constraint of mobile hardware resource and wireless network bandwidth, a viewpoint-predicting-based remote rendering on mobile devices using multiple depth images is proposed. First, a reference viewpoint selection mechanism is given based on user interactions at present, and then guide the generation of corresponding depth images. Secondly, an edge based depth map reconstruction method is brought into encoding and decoding in order to furthermore improve rendering efficiency meanwhile reduce total transmit data on the network. Final synthesis result is produced with the received multiple depth images using 3D image warping. The experimental results show that it not only rises the rendering quality but also reduces the quantity of transmit data, hence meets the requirement of real-time interaction on mobile devices.