摘要
针对应用中非下采样轮廓波冗余度过高、变换运行时间长的缺点,提出一种循环平移抗混叠轮廓波(NACT)图像融合方法。该方法利用NACT对图像进行分解,同时使用循环平移的方法提高NACT的平移不变性,有效去除融合图像在奇异点处产生的虚假信息。实验结果表明,该方法能够改进图像的融合性能,在客观指标上较NACT、NSCT等方法有所提高,视觉效果有明显改进。
Aiming at the weakness of high redundancy and long run time existing in nonsubsampled contourlet transformation, the Non-Aliasing Contourlet Transform（NACT） for image fusion using cycle spinning is proposed. NACT is utilized as the multi-scale transformation to decompose the original images into subbands and cycle spinning is applied to improve the NACT translation invariance and overcome false information of fusion image. Experimental results show that the method is effective in improving the performance of image fusion and is better than the method using contourlet or NSCT in objective index, meanwhile visual effect is significantly improved.
出处
《计算机工程》
CAS
CSCD
北大核心
2011年第4期241-243,共3页
Computer Engineering
关键词
循环平移
抗混叠轮廓波
融合规则
方向矢量
空间频率
cycle spinning
Non-Aliasing Contourlet Transform（NACT）
fusion rule
direction vector
spatial frequency
