摘要
多遥感图像融合的高频部分在区域处理时易出现边界歧异,造成图像失真的问题。为了提高图像视觉效果和特征识别,提出一种新的分块区域遥感图像融合方法。首先采用多小波分解把图像分解成不同尺度的高频和低频两部分,然后在不同频带选择不同的融合规则,在高频处理部分,采用分块区域处理,取区域绝对值较大的作为融合后的多小波系数,低频采用加权平均算法,最后经多小波重构得到最终融合图像。仿真结果表明,改进方法能够完好地保留遥感图像的原始信息,融合效果良好。
A new remote sensing image fusion method combined with the features of multi-wavelet was proposed in this paper,which avoids boundary interferences caused by regional processing in the high-frequency part.Firstly,the original images were decomposed to different scales of high-frequency and low frequency.Secondly,the image fusion was then implemented in different bands.The regional blocking is adopted in high-frequency,the larger absolute value of the region was taken as fusion coefficients,and the Weighted Arithmetic Average algorithm was adopted in low-frequency.Finally,the final fused image was obtained using multi-wavelet reconstruction.The results of experiment show that this fusion method can perfectly remain the original information and its results are much more effective.
出处
《计算机仿真》
CSCD
北大核心
2011年第12期249-252,共4页
Computer Simulation
关键词
遥感图像
多小波
图像融合
分块区域
Remote sensing image
Multi-wavelet
Image fusion
Regional blocking
