欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
采用SCDPT变换对图像进行多尺度、多方向分解,得到图像不同尺度、不同方向的频带系数.然后对低频子带系数采取基于结构相似性（SSIM）、区域能量和区域平均梯度的融合规则,对方向子带系数采取基于SSIM和区域方差的融合规则.最后通过SCDPT逆变换得到融合图像.采用信息熵、平均梯度、互信息、边缘强度、均值等作为客观评价指标,实验结果表明,相对于小波变换、拉普拉斯金字塔变换、梯度金字塔变换,所提出的算法能够充分提取图像特征,具有更灵活的方向性和平移不变性,并且能够准确捕获图像轮廓特征信息和纹理细节信息.融合结果优于大部分基于其他多尺度变换的图像融合算法.
英文摘要：
Shiftable complex directional pyramid transform（SCDPT） is used to decompose source images at each scale and direction to obtain low-pass sub-band coefficients and band-pass directional sub-band coefficients.Low-pass sub-band coefficients are fused w ith the rule based on structural similarity（SSIM）,regional energy,and regional average gradient.Directional band-pass sub-band coefficients are fused w ith the rule based on SSIM and regional variance.Finally,fused image is obtained by SCDPT inverse transform.Information entropy,average gradient,mutual information,edge strength,and mean are adopted as objective evaluations.Experimental results show that,compared w ith the w avelet transform,Laplacian pyramid transform and gradient pyramid transform,the proposed algorithm can not only fully extract image features w ith more flexible directivity and shift invariance,but also can accurately capture the image information of the contour feature and texture details.Fusion result is better than those of the most other algorithms based on multi-scale transformation.
同期刊论文项目
同项目期刊论文
Image Fusion based on Nonsubsampled Contourlet Transform and Saliency-Motivated Pulse Coupled Neural
期刊信息
位置：