Calibration Of Fisheye Camera Images And The Application In Video Enhancement Scene

With the development of virtual and real fusion technology,the use of video images to enhance three-dimensional scenes makes static models reflect real-world changes in real time,which is of great significance for security and intelligent transportation.Compared with ordinary cameras,the fisheye camera has a larger angle of view,and the collected fisheye image information is also richer.However,due to the very large distortion in fisheye video,the current video enhancement scene algorithm is difficult to accurately fit the fisheye video with the three-dimensional model,so this paper improves the existing video enhancement scene algorithm.The accurate matching of the fisheye video and the three-dimensional model is realized.Before performing video enhancement scenes,it is necessary to calibrate the fisheye camera.Currently camera calibration methods include calibration methods based on calibration plates,camera self-calibration methods,etc.,but these methods often require the use of certain calibration objects such as black and white checkerboard To achieve calibration,or need to remove the camera in the laboratory environment to achieve calibration,in order to solve these problems,This paper proposes a fisheye camera calibration method based on genetic algorithm.The research content and results of this article mainly include:(1)A fisheye camera calibration algorithm based on genetic algorithm.The algorithm obtains the coordinates of the principal point through the circle detection algorithm,and then uses the genetic algorithm to solve the distortion coefficient of the fisheye image.Through the demonstration,it is found that the spherical projection model of the fisheye can be converted into Perspective projection model,so turn the fisheye image into a normal picture without distortion,and then use the world coordinates of the feature points and the image coordinates to calibrate according to the two-step method.Although this method is lower in accuracy than the fisheye camera calibration method based on the calibration plate,this method can achieve the calibration of the fisheye camera without the aid of the calibration plate,and the application scenarios are more extensive.(2)The fisheye video enhanced scene algorithm,using the fisheye camera calibration parameters to construct a virtual camera in the Open GL three-dimensional scene,and then drawing on the idea of the Shadow Map algorithm to map the fisheye video texture.Before texture replacement in the shader,the image is distortion-corrected,the image coordinates are back-projected to obtain world coordinates,and then the world coordinates are re-projected according to the spherical projection model,and distortion correction is added during re-projection to obtain the corrected For image coordinates,use the corrected coordinates to extract the corresponding RGB from the fisheye image for texture replacement,so that the fisheye image fits the model more accurately.