The 3D Image Processing To Accelerate Based On The Technology Of GPGPU

General Purpose graphics processor(GPGPU) is new trend in the development of the current graphics processors(GPU). Due to the particularity of the GPU, make its architecture suitable for graphic images of a large number of floating point arithmetic, so the current GPU has numerous hardcore computing units. To get so many computing unit of computing power, one some difficult to use the CPU intensive computing will try to convert image problem actually before GPGPU appear, and then use this abstract graphics interface of GPU programming, to complete some common calculation. The main content of the research and the main innovative achievements summarized as follows:1、By GLSL shader programming of GPU to realize 3D image processing, multiple viewpoints parallel computing processing speed.2、Based on CUDA-OSG 3D image acceleration method, using CUDA parallel computing to finish virtual view of graph generating and hole filling and occlusion processing, study how to control the image rendering and mapping, and pixel processing order and depth gradient filter, to eliminate the hole and occlusion in the GPU parallel computing problems.3、Put forward a kind of suitable for large size LCD screen processing structure, using a client-server approach can achieve auto-stereo real-time interactive processing of electronic sand table.4、In order to better show the GPGPU in image processing speed advantage, based on cube collision simulation scenarios are studied,including the parallelization of the simulation of rigid body, particles,broadband test to do the relevant research.This article research results enrich the GPGPU in auto-stereo display acceleration and related theory, physics engine technology for GPGPU in auto-stereo display and the application of physics engine to explore the technical scheme, for high performance, low power consumption, real time, large capacity GPGPU provides theory basis for image processing speed, is expected in the next GPGPU in virtual reality and acceleration is widely used in image processing.