Correction And Mosaic Algorithm Of Panoramic Image

Recent years, with the continuous development of urban economy, private cars are increasing day by day, which makes urban road congestion become more and more serious and bring great impact on urban traffic. Among which the overcrowding road cars lead to drivers’dead zones while driving and parking, causing traffic accidents frequently, which make threats to the life security of drivers and pedestrians on the road. Aiming at this phenomenon, this paper puts forward a set of on-board panoramic display system, also known as 360-degree panoramic display system. The system works by installing cameras around the body, gathering the images around, collecting four images splicing into a picture under the direction of aerial view by means of image transformation, which enables the driver get to know the surroundings in four directions on the screen in cab, and assists driving safely and quickly, reduces traffic accidents caused by blind zones.The thesis is mainly divided into three parts:the first part mainly introduces detailed information of the imaging lens used by the system and based on which studies the imaging process in a panoramic view of the entire system. The second part mainly introduces the algorithm of image processing, after image stitching of the four pictures, converting it into an aerial view from the car in the look-down direction. Simply realizes fisheye image distortion correction with high-efficiency. While the inverse projection transformation adopts the analysis of image in three-dimensional geometric. The last part is image stitching, which is also the most core part of the system. This part consists of five small parts:image registration, modeling, image transformation conversation, image fusion and image stitching algorithms comparison. The image registration uses special calibration experiment environment, which made the whole system image transformation greatly simplified. In the article, considering the real time on-board system, and improving the system running rate, the adoption of the method of the matrix is derived, and four deputy fisheye images can be collected, joining together into a panoramic image directly through the matrix of point to point mode, which greatly improves the speed of the system and the system running time and makes real-time display environment possible. Finally, in order to keep the display image in the joint continuing, there still needs the final image fusion algorithm of the image processing.The third part mainly introduces the hardware environment of the whole vehicle panoramic display system and the debugging of hardware environment effect. In the introduction of hardware platform part, mainly introduces overall design train of thought of the on-board panoramic display system as well as the design results of the final design.