CAJ | 학술논문

针对像素级多传感器图像的融合问题,设计了一种基于梯度变换的多传感器图像融合算法。该算法利用金字塔变换得到图像的多分辨序列,在相应层次的金字塔图像上进行多方向分解,采用基于梯度变换的融合方法以获取最终的图像。从目视效果来看,采用该算法进行图像融合,能将2幅多传感器图像融合成1幅信息量饱满的单一图像。实验结果表明,该算法与同类其他算法相比较,标准差提高了0.05%～13.66%,偏差度降低了10.33%～48.49%,联合熵提高了0.36%～7.22%。
침대상소급다전감기도상적융합문제,설계료일충기우제도변환적다전감기도상융합산법。해산법이용금자탑변환득도도상적다분변서렬,재상응층차적금자탑도상상진행다방향분해,채용기우제도변환적융합방법이획취최종적도상。종목시효과래간,채용해산법진행도상융합,능장2폭다전감기도상융합성1폭신식량포만적단일도상。실험결과표명,해산법여동류기타산법상비교,표준차제고료0.05%～13.66%,편차도강저료10.33%～48.49%,연합적제고료0.36%～7.22%。
Focusing on the fusion problem of the multi-sensor images for pixel level, the multi-focus image fusion algorithm based on regional energy was designed. By using muhiresolution image sequences obtained from the pyramid transform, the source images were decomposed on the corresponding level of multi-direction, and the source images were fused on each corresponding levels with the gradient transform method, then the fused image was achieved through inverse pyramid transform. From the visual effect, the algorithm can fuse into a full information content image from two-sensor im- age. Experimental results and comparisons with three traditional algorithms demonstrate that the designed algorithm is superior with 0.05% to 13.66% improvement of standard deviation, 10.33% to 48.49% reduction in the deviation degree and 0.36% to 7.22% enhancement in union entropy.