Car Panorama System With Multi-source Based On Image Fusion

With the development of science and technology, the cars have become an important vehicle. At the same time, the road was crowded by the increase number of cars, and the kinds of traffic accidents were frequently occurred to bring some trouble for the people. Then the car panorama system with multi-source was produced to avoid unnecessary accident. And the images of around the car were divided to provide the driver. The panoramic system can't be stitching into a panorama on the market, or the poor effect of panorama images was divided to influence people's visual effect. So the clear images of car panoramic and simulation experiment were researched.First of all, the method of approximate estimation for accurate positioning of fish-eye lens was proposed to solve the installation location was not precise in panoramic parking and improves the precision of image mosaic. The four fish-eye lenses were installed around of the car. And the appropriate space coordinate system was established by regarding the car as the main body. Then, the relationship between the ground and installation parameters of the lens was considered. The gathering area of the image was projected onto the xoy coordinate system to transform the intersecting ellipsoid with overlap area in the xoy coordinate system. And the area of different overlap region can be calculated by using the approximate estimation method, thus the optimal tilt angle of fish-eye lens was obtained.In addition, the collect images were divided to the panoramic image. The method of nonlinear equation for distortion correction was proposed to improve the correction accuracy of fish-eye image, and the fish-eye images were transformed into a square image on the basis of vertical and horizontal coordinates. Then the optimization method of the Harris corner detection for feature point extraction was proposed to reduce the time of feature point extraction. And the optimization method of image fusion for the overlap region was proposed to improve image clarity, and a clear and precise panoramic image was obtained.At last, the proposed methods were verified by the experiments of simulation model. The four cameras were installation on the model, and the image around the car collected. Base on this, the MATLAB GUI was used to design interface of simulation experiment, and the car panorama system with multi-source was designed.