Research And Implementation Of Vehicle Panoramic Image Mosaic Algorithm

With the rapid development of China's economy,cars are widely used in thousands of households.As the main means of transportation in the world today,the car is not only convenient for people,but also a variety of traffic accidents.Analysis of the main causes of automobile traffic accidents,a variety of statistical data show that the accident occurred in the blind area of the car.The blind area is easy to cause the driver to make the wrong judgment to the environment of the automobile,thus causing the pedestrian collision,the vehicle scratches and so on.Aiming at the problem of blind spot in vehicle environment,the solution of vehicle panoramic image is presented.The vehicle panoramic image is provided to the driver with a panoramic view of the car without blind spots.The driver can understand the environmental information in the blind area by observing the view screen,so as to avoid the occurrence of traffic accidents.In the vehicle environment,the installation of four fish eye camera to collect the image information around the car,the video capture card will be sent to the image of the four image information into the processor for panorama synthesis,the final result is displayed on the LCD screen.In the process of panorama synthesis,firstly,the fish eye image is corrected by distortion,and then the view is adjusted to get the top view.Finally,the registration and fusion of the four directions are carried out.After analyzing the various steps of the panoramic image mosaic algorithm,the main contents of this paper are as follows:(1)The fisheye image distortion correction step,based on the imaging model calibration algorithm with high accuracy,isometric projection model comprehensive common good correction effect,orthogonal projection model of edge correction good effect,the use of a fish eye image correction algorithm based on weighted average imaging model,the image distortion at the edge and interior have good correction effect.(2)On the basis of the simple and effective direct linear transformation algorithm,the transformation process does not need to know the external parameters of the camera,but only four pairs of vertex coordinates in the calibration space.In view of the fact that the observation error caused by the fixed point measurement error in the actual calibration process is not good,this paper adopts the modified direct linear transformation algorithm,which makes the effect of the top view transformation more accurate.(3)In the step of image registration transformation,this paper arranged on the experimental environment,in the car around a circle of white box line,in line with rich edge information of the diagonal area as a template,found in the search map registration sub map location.In this paper,we use the edge template based registration algorithm to reduce the amount of computation while using the edge registration information to obtain accurate registration results.In the step of image fusion,this paper uses the method of weighted fusion of trigonometric functions to get a better fusion effect.At the end of this paper,we discuss the application of the simulation algorithm in PC to the automotive environment.Finally,we design a set of panoramic vision system software.The experimental results show that the panoramic image mosaic algorithm designed in this paper can meet the requirements of real-time and effectiveness of the system.