基于角点检测的彩色图像拼接技术

Color image mosaic technology based on corner detection and affine transform model

针对彩色图像的拼接问题,提出了一种基于角点检测和仿射变换模型的图像配准算法。利用Harris算子在彩色图像的HSV空间的亮度分量V中检测角点,然后引入基于灰度相关交叉匹配和简化RANSAC算法的两级约束机制确定同名角点,并计算图像仿射变换模型参数,实现图像配准。同时结合人眼的视觉非线性给出了一种分段加权函数,通过对饱和度分量S和亮度分量V的处理,消除明显的拼接缝,实现色彩和亮度的平稳过渡。实验结果表明,该算法具有良好的拼接精度和拼接缝消除效果。

An image registration method based on the comer detection and affine transform model is proposed. This method uses the Harris operator to detect the comers in the HSV space of the color images. A two-level restricted tactic based on cross match and simplified RANSAC algorithm to gain the homonymy comer points is introduced. A segment-weighted function based nonlinear characteristic of human vision is given to eliminate the mosaic linecaused by the difference ofthe image color. The experiment indicates that this algorithm has good mosaic precision and nice fusion ofpixel color.