投影模型下全方位视觉图像畸变校正算法仿真
Simulation of Distortion Correction Algorithm for Omnidirectional Vision Image under Projection Model
孙海霞 1胡永1
作者信息
- 1. 西藏民族大学信息工程学院,陕西 咸阳 712082
- 折叠
摘要
为提高全方位视觉图像内目标清晰度,强化边缘细节特征,调整畸变的几何比例结构,提出一种投影模型下图像畸变校正算法.构建图像畸变模型,明确畸变与原始图像间非线性变化关系,通过正交投影变换构建透视投影模型和平行透视投影模型,将平行透视投影近似看作为透视投影的一阶结果,计算二者仿射近似误差值,最后将畸变校正问题定义为畸变与理想点的位姿重合,令位姿校正转换为重合最小距离求解,使用齐次坐标获得校正的旋转和平移量,实现图像畸变校正.仿真结果表明,所提算法校正后图像边缘清晰,细节部分无锯齿现象,说明上述算法能够精准还原目标特征,具有较高的可应用价值.
Abstract
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.
关键词
投影模型/视觉图像/畸变校正/弱透视投影/平行透视投影Key words
Projection model/Vision image/Distortion correction/Weak perspective projection/Parallel perspective projection引用本文复制引用
出版年
2024
NETL
NSTL
万方数据