摘要
为了进一步提高图像配准的运算效率、匹配正确率及配准精度,提出了一种利用双树复小波变换和加速鲁棒特征(speeded up robust features,SURF)的图像配准算法。首先利用双树复小波变换将参考图像和待配准图像分解为低频部分和高频部分,选取其对应的低频部分作为SURF算法的输入图像,得到两者的粗匹配结果;然后通过随机抽样一致(random sample consensus,RANSAC)算法对粗匹配点对进行提纯,剔除误匹配点对,解决了SURF算法存在较多错误匹配点对的问题,同时计算出最佳匹配的变换模型参数;最后根据该变换模型参数对待配准图像进行几何变换,经双线性插值确定灰度,完成图像的配准。大量实验结果表明,与尺度不变特征变换(scale invariant feature transform,SIFT)算法和SURF算法相比,所提算法的运算速度更快,匹配正确率和配准精度更高,同时在抗噪声、抗旋转及抗亮度变化性能方面更加优越。
To further improve the operation efficiency,correct matching rate and registration accuracy of image registration,an image registration algorithm based on dual tree complex wavelet transform and speeded up robust features （SURF）is proposed.Firstly,the standard image and the image to be registered are decom-posed into the low and high frequency parts by dual tree complex wavelet transform.The selected corresponding low frequency parts serve as the input image of the SURF algorithm,to obtain the coarse matching results. Then the coarse matching point pairs are purified and the mismatching point pairs are eliminated by the random sample consensus （RANSAC）algorithm.Thus the problem of more mismatching point pairs caused by the SURF algorithm is solved.Meanwhile,the transform model parameters of the optimal matching are calculated. Finally,the geometric transformation of the image to be registered is performed according to the transform mod-el parameters,the gray level is determined by bilinear interpolation,and the image registration is completed.A large number of experimental results show that the proposed algorithm has a higher calculation speed,correct matching rate and registration accuracy than the scale invariant feature transform （SIFT）algorithm and the SURF algorithm.It also performs better in resisting noise,rotation and brightness change.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2014年第5期997-1003,共7页
Systems Engineering and Electronics
基金
国家自然科学基金(60872065)
农业部农业科研杰出科技人才基金
农业部农业信息技术重点实验室开放基金(2013001)
江西省数字国土重点实验室开放基金(DLLJ201412)
江苏高校优势学科建设工程资助课题
关键词
图像配准
双树复小波变换
加速鲁棒特征
随机抽样一致算法
image registration
dual tree complex wavelet transform
speeded up robust features （SURF）
random sample consensus （RANSAC） algorithm
