一种抑制噪声的多聚焦图像融合算法

摘要

Noise pollution is often introduced during image acquisition and transmission process. Existing image fusion methods can not identify meaningful image features from noise. A new method which combines image block segmentation and multiscale products of Stationary Wavelet Transform is proposed for fusion of noisy multi-focus images. Theoretical analysis and experimental results show that the algorithm can better protect edge features and useful image information while restraining noise. The better visual effects are gained compared with the traditional fusion methods, while reducing the computation.%图像在获取和传输过程中,通常会受到噪声的污染.现有的融合算法不能很好地区分噪声和视觉上有意义的特征信息,针对此问题提出了一种结合图像块分割与平稳小波域多尺度积变换的多聚焦图像融合算法.理论分析和实验结果表明,算法在抑制噪声的同时很好地保护了边缘特征信息和图像的有用信息,得到了比传统融合方法更优的视觉效果,可靠性高,实时性好.