广角图像畸变校正算法的研究与实现

摘要

为了校正广角图像的非线性畸变，提出一种新的数字校正方法来消除畸变。首先利用网格模板校正的方法，根据畸变图与理想图对应像素点的映射关系，得到畸变图像点在x轴和y轴方向上的偏移量。然后采用三次B插值函数对曲面插值，得到畸变像素点的偏移量曲面，由偏移量曲面和畸变点的坐标实现各像素点的坐标变换。最后通过双线性插值法完成灰度重建得到无畸变的图像，从而实现对广角图像的校正。为了测试该算法的速度性能和可靠性等指标，在DSP平台上运行此算法。实验结果表明该算法能够对广角畸变图像进行快速有效的校正。%In order to correct nonlinear distortion of wide-angle images, we propose a new digital correction method to eliminate the distortion.First, it makes use of grid template correction method to obtain the offset of distorted image points along X-and Y-axes directions according to the mapping relation between the corresponding pixels of distortion image and ideal figure.Then, it uses the cubic B-spline interpolation function to interpolate the surface, and gets the offset surface of the distorted pixels.According to the offset surface and the coordinate of distorted points, it realises the transformation of coordinate of each pixel.At last, a perfect image without distortion is obtained by the completion of grayscale reconstruction with bilinear interpolation, so as to achieve the correction on wide-angle images.In order to test the speed performance and the reliability of the algorithm, we run it on DSP platform.The experimental results show that this algorithm can make fast and effective correction on wide-angle distortion images.