CAJ | 학술논문

针对基于主元分析的图像融合算法存在结构利用率低、光谱信息损失多的缺点，同时考虑到色彩图像融合时空间变换产生的色彩畸变以及RGB色彩空间各通道间的强相关性，提出一种基于改进双向二维主元分析的图像融合算法。针对RGB色彩图像的结构特点，以待融合图像行、列方向的RGB分量作为基元进行二维主元分析得到各级主元，采用线性权重分配方法对待融合图像进行重构，依照重构图像第一主元的结构特性进行主元替换后，经加权逆变换得到融合图像。为验证算法的有效性，选取校园近红外光谱图像与对应的清晰彩色图像，以及彩色可见光图像与对应的红外图像进行实验，实验结果表明使用文中方法得到的融合图像可取得理想的融合指标和较好的空间分辨率。
침대기우주원분석적도상융합산법존재결구이용솔저、광보신식손실다적결점，동시고필도색채도상융합시공간변환산생적색채기변이급RGB색채공간각통도간적강상관성，제출일충기우개진쌍향이유주원분석적도상융합산법。침대RGB색채도상적결구특점，이대융합도상행、렬방향적RGB분량작위기원진행이유주원분석득도각급주원，채용선성권중분배방법대대융합도상진행중구，의조중구도상제일주원적결구특성진행주원체환후，경가권역변환득도융합도상。위험증산법적유효성，선취교완근홍외광보도상여대응적청석채색도상，이급채색가견광도상여대응적홍외도상진행실험，실험결과표명사용문중방법득도적융합도상가취득이상적융합지표화교호적공간분변솔。
The color transform during image fusion produces color distortion; the existing image fusion algorithms based on principal component analysis ( PCA) do not fully utilize image structures, thus losing much spectral infor-mation. Hence, we propose the color image fusion algorithm mentioned in the title. We use the RGB components of a color image in its row and column directions as the base components to do two-dimensional and two-directional PCA and reconstruct the color images to be fused, thus retaining its structural information. The color image is fused with the linearly weighted and reverse transform based on the eigenvalues of a covariance matrix. The color image thus fused retains high-resolution information and energy information of the infrared thermal image. To verify the ef-fectiveness of our color image fusion algorithm, we fuse a blurred remote sensing color image and its corresponding clear remote sensing gray image and then fuse a visual color image and its corresponding infrared image. The fusion results, given in Figs. 3 through 5 and Tables 1, 2 and 3, and their analysis show preliminarily that our image fu-sion algorithm can greatly reduce the color distortion and obtain satisfactory fusion effects.