A Fast Correction Method for Fish-Eye Image Based on Inverse Latitude-Longitude Projection
秀华 章 · Journal of Image and Signal Processing · 2018
针对各种不同相机拍摄的鱼眼图像，提出了一种基于模板矩阵的鱼眼镜头畸变图像快速校正方法，先将鱼眼图像和目标图像分别划分为上下左右四个像素区域。利用逆向经纬映射方法，计算目标图像其中一个像素区域上的每一个坐标点在鱼眼图像上对应点的坐标，并将计算得到的坐标值保存在一个坐标模板矩阵中，建立其它三个区域上的像素坐标点与第一个像素区域上坐标点之间的对应关系，可直接从模板矩阵得到目标图像其它三个像素区域在鱼眼图像中对应点坐标。对不同相机拍摄的各种不同分辨率鱼眼图像进行处理，实验结果表明提出的校正方法耗时少，速度快，具有较强的实用性。 For many fish-eye images captured by different cameras, a fast correction method for distorted fish-eye image based on model matrix is proposed. Firstly the fish-eye image and object image are divided into four areas: up, down, left and right. Inverse latitude-longitude projection method is used to compute the coordinate of the first area in the fish-eye image corresponding to that in the object image, and the computed coordinate values are saved in a model matrix. The coordinate relationship between the first area and the three other areas is constructed; thus the coordinate values of three other areas can be derived from the first area. Many different sizes of fish-eye images are processed by the proposed method. The results show the speediness and good practicability of the proposed method.