期刊论文

Multi-focus Image Fusion Based on Block DCT Transform

中文

小型微型计算机系统

1000-1220

2016

37

2

321-326

国家自然科学基金项目（61304208）资助；

本校为第一机构

提出一个简单而有效的多聚焦图像融合方法,即基于分块DCT变换的多聚焦图像融合,首先对图像进行8* 8分块,然后分别对每个8* 8块进行DCT变换,接着计算归一化的交流系数的绝对值的均值(|AC | _mean),其中融合规则是选取| AC| _mean大的8* 8图像块进行融合,最后通过一致性评估降低块效应. 实验中本文算法与DWT、SIDWT、DCT + Variance、DCT+ SF算法进行了对比,通过实验证明了本文算法图像融合的效果与质量较佳,边缘细节信息较显著,图像融合取得了较好的客观评价与主观视觉效果,并且本文算法涉及的计算复杂度低,可为一些有限能源的图像处理设备(如用于临时监控的无线视觉传感器实时系统)提供技术支持...

This paper presents a simple but efficient way of multi-focus image fusion,namely,the block DCT-based multi-focus image fusion,in which the image is first 8* 8 partitioned,and then DCT transformation is done to each 8* 8 block,and then the computes the mean of the absolute value of the normalized AC coefficient(|AC | _mean) is computed. The fusion rule is to fuse the 8* 8 blocks whose |AC | _mean value is bigger. Finally,by Conformity Assessment to reduce the block effect. In the experiments the proposed method is compared with the other methods such as DWT,SIDWT,DCT + Variance,DCT + SF. The e...