基于NSST图像特征的多聚焦图像融合方法

针对广泛存在于多聚焦图像融合方法中的局部图像细节不清晰的现状,提出一种基于非下采样剪切波变换(NSST)图像特征的多聚焦图像融合方法。利用NSST对待融合源图像进行多尺度、多方向稀疏分解,分别获取低频和一系列高频子带图像。通过空间频率和局部能量确定融合后的低频子带系数,利用边缘检测算子直接获取高频子带图像中的细节和边缘信息,并采取NSST反变换得到最终融合结果图像。仿真实验结果表明,同现有的几种经典算法相比,本文提出的方法获得的结果图像拥有更清晰的视觉效果、更理想的客观指标效果以及更高的算法运行效率。

Fusion Technique for Multi-focus Images Based on Non-Subsampled Shearlet Transform and Image Characteristics

Faced with the current situation that the visual information of local image details is blurred in many fusion techniques for multi-focus images, a novel fusion technique for multi-focus images based on non-subsampled shearlet transform (NSST) and image characteristics is proposed. The low-frequency and s series of high-frequency sub-band images are obtained by conducting the multi-scale and multi-directional decompositions of the source images via NSST. The spatial frequency and local energy are utilized to determine the fused coefficients of low-frequency sub-band. The information of the details and edges existing in the high-frequency sub-band images can be directly achieved by the edge detection operator. The final fused image can be obtained via inverse NSST. Simulation experimental results demonstrate that, compared with several current typical algorithms, the fused image based on the proposed technique has much clearer visual performance, more reasonable objective metric results and much higher functional efficiency.