OpenGL Acceleration Technology For Panoramic Image Mosaic

With the development of automotive electronic technology and image processing technology,vehicle assisted intelligent driving system is more and more used in vehicles.As the most important part of the assisted driving system,panoramic image can provide drivers with a wide field of vision and greatly reduce the blind area of vision,which has been paid more and more attention by people.In recent years,national car ownership has increased year by year,and traffic accidents have continued to increase.The promotion of panoramic look will help to reduce accidents.However,due to the certain requirements for computing power of panoramic view,at present,the domestic market mostly adopts foreign vehicle chips with high computing power,and the price is high,which is difficult to popularize in low and medium-sized models.Thesis mainly studies the high-performance implementation method of the panoramic surround view system on the domestic low-computing platform.Firstly,the distortion model and imaging principle of fisheye camera are studied,and the calibration method of 2D and 3D panoramic view is obtained through experiments.Then,the domestic chip platform is researched,and its embedded development environment is built.On this platform,the V4L2 framework is used to realize the acquisition of fisheye camera data,and the processing speed bottleneck caused by the copying of video data is reduced by DMA transmission.Then,the working principle of OpenGL is studied,and the acceleration of 2D panoramic view by GPU is realized through shader programming.Finally,an algorithm is used to create a 3D bowl-shaped model,and through the coordinate system transformation of OpenGL,the imaging principle of the fisheye camera is simulated,and the texture mapping of the 3D panorama is realized.Experiments show that after OpenGL acceleration,the 2D panoramic view can reach 30 frames per second,the 3D panoramic view can reach 25 frames per second,the system resource occupancy rate does not exceed 50%,and the display effect of the mainstream panoramic view in the market can be achieved,satisfying the panoramic view.Looking around is based on the mass production demand of domestic chips.Thesis completed deploying the algorithm in embedded system with a domestic chip development board,and the real vehicle test is carried out.The test shows that after accelerated by OpenGL,the 2D panoramic look around can reach 30 frames / s,the 3D panoramic look around can reach 25 frames / s,the system resource occupancy rate is no more than 50%,and can achieve the display effect of mainstream panoramic look around in the market,meeting the engineering application requirements of panoramic look around based on domestic chips.