摘要
针对光照方向、强度、色彩或摄像头参数等因素的变化导致待配准彩色图像对间存在复杂的色彩变化,从而严重影响彩色图像配准效果的问题,提出了一种对彩色变化不敏感的彩色图像配准方法.该方法首先根据彩色图像间的von Kries彩色变换模型,建立图像的彩色不变量空间,在此空间利用尺度不变量特征转换(SIFT)算法完成特征点的检测;为加快运算速度,采用2阶局部差分模型(LDP)方法对特征点进行描述,并完成彩色图像特征间的初匹配;再进一步利用RANSAC算法消除误匹配特征点对,得到最终的匹配特征点对.实验结果表明,该彩色图像配准方法与其他算法相比,可以快速、准确地获得彩色图像间的映射关系,因此更加适合存在色彩变化的彩色图像的配准.
Complex illumination change, caused by changes of direction, intensity and color of illumination, or camera parameters, exists between color image pair, and seriously impairs the effect of color image registration. A color image registration algorithm which is insensitive to the illumina tion change is proposed to solve the problem. The algorithm transforms the original RGB color image space to a color image invariant space by using the yon Kries color transform model. Then key-points are detected using the scale invariant feature transform (SIFT) algorithm in the color invariant space, and are described using the second-order local derivative pattern (LDP) method to accelerate matching speed and to complete elementary matching. The RANSAC algorithm is employed to discard mismatching feature pairs and to obtain the matching feature sets. Experimental results indicate that the proposed algorithm generates mapping relation more correctly and quickly than other color image registration methods. Hence, the proposed color image registration algorithm is suitable for color images in which there exists color illumination transform.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2011年第10期30-37,共8页
Journal of Xi'an Jiaotong University
基金
黑龙江省自然科学基金资助项目(AF200921)
关键词
彩色图像配准
彩色不变量特征
尺度不变量特征转换算法
局部差分模型
color image registration
color invariant feature
scale invariant feature transform (SIFT)
local derivative pattern (LDP)