Simulation of Distortion Correction Algorithm for Omnidirectional Vision Image under Projection Model
In order to improve the definition of object in omnidirectional vision images,strengthen the edge details,and adjust the geometric structure of distortion,this paper presented an algorithm of correcting image distortion based on projection model.Firstly,we built an image distortion model,and defined the nonlinear relationship between distortion and original image.Secondly,we built a perspective projection model and a parallel perspective projection model through orthogonal projection transformation,and regarded the parallel perspective projection as the first-order result of perspective projection approximatively.Thirdly,we calculated the affine approximation error value of the two.Finally,we defined the distortion correction problem as the coincidence between distortion and ideal point,so that the pose correction could be converted into the solution of minimum coincidence distance.Meanwhile,we used the homo-geneous coordinates to obtain the corrected amount of rotation and translation,thus realizing the correction of image distortion.Simulation results show that the proposed algorithm has clear image edges without aliasing in details after correction,indicating that the algorithm can accurately restore the target features and has high application value.
NETL
NSTL
万方数据
