摘要
针对普通光学系统对图像处理算法实时性要求高及噪声环境的特点,提出一种基于拉普拉斯塔型分解伪逆重构的抗噪声图像融合算法.采用正交小波分解构建图像数据金字塔结构,重构过程采用一种新的伪逆重构算法.新算法可抑制压缩过程量化、阈值处理等带来的系统噪声及加性白噪声,算法复杂度低.可获得比原拉普拉斯金字塔变换法较优的多聚焦图像融合效果,并具有抗噪声能力.
In view of the more real-time requirements of the image processing algorithm for ordinary optics and the particularity of noise condition, this paper proposes an antinoise algorithm for image fusion, which is based on Laplacian pyramid decomposition and pseudo inverse. For the algorithm, orthogonal wavelets are used for image decomposition and the pyramid data structure, and at the reconstruction stage, a pseudo inverse is used. The new algorithm can restrain the additive noise and system noise from thresholding and quantization process of compression, and it is of lower complexity. Experimental results show that the new algorithm can obtain a better effect than the method based on the usual Laplacian pyramid transform for the multi-focus image fusion, and has antinoise properties.
出处
《西安电子科技大学学报》
EI
CAS
CSCD
北大核心
2011年第3期63-68,共6页
Journal of Xidian University
基金
国家自然科学基金资助项目(60802077
61003196)
