摘要
提出了图像特征塔集的概念,给出了一种基于特征塔集的融合方法.算法主要包含两部分,子图像的融合及对应残差图像的融合.图像的大部分能量和信息包含在子图像中,因此,在图像金字塔的子图像融合过程中,特征塔集的构造由局部能量和局部熵组成.考虑到残差金字塔包含了原始图像的边缘或高频成分,因此采用梯度作为其融合特征.最终的融合结果,可利用金字塔反变换获得.由于将残差金字塔也引入到融合算法的设计过程中,因此融合的精度得到一定的提高,另外,采取简单的金字塔可以有效地控制算法的复杂度.实验结果表明了算法的合理性和有效性.
With the development of remote sensing, it is not difficult to capture different images with different resolutions, waves band and sensors for certain same scene. Image fusion is an important measure to integrate the different advantages of different images of a same scene. Applying the idea of multi-resolution analysis and combining the strategy of multiple feature fusion, this paper proposed a concept, image character pyramid set ICPS and described an image fusion algorithm based on PSIC. The novel algorithm consist of two parts. One is the fusion for the sub-images; the other the fusion for the residual images. Because most of the energy and information are contained in sub-image, so the PSIC is constructed by local energy and entropy in the fusion procedure of sub-images. The edge or high frequency component of the original image being contained in the residual pyramid, the gradient future is used for the fusion of residual pyramid, the gradient future is used for the fusion of residual pyramid set of image. The final fusion result is acquired by inverse transform of pyramid set of image. Because the residual pyramid is introduced into the fusion procedure, the fusion accuracy is improved. At the same time, the complexity of the algorithm is effectively controlled by the usage of simple pyramid. The experiments illustrated the method.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2005年第7期14-16,共3页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(60272099)
航天科技创新基金资助项目(20030956)
湖北省自然科学基金资助项目(2004ABA501).
