基于多孔小波和模糊区域特征的图像融合方法

Multi-source Image Fusion Scheme Based on àtrous Wavelet and Fuzzy Region Feature

提出了基于模糊区域特征的图像融合方法,该方法是在多孔小波(àtrouswavelet)多分辨率分析的基础上,根据图像每层的低频分量进行K-mean,将低频图像分解为重要区域、次重要区域以及背景区域;图像各个区域针对不同属性模糊化,根据每个区域各自的模糊隶属度来确定每一部分区域的融合策略,最终得到融合图像的多尺度表示,再对其进行多分辨率反变换就可以得到图像的融合结果。实验表明,该文提出的基于模糊区域特征的图像融合方法具有较好的融合结果。

An image fusion scheme based on fuzzy region feature is proposed using discrete wavelet frame multi-resolution analysis. Every level low frequency band is applied to segment image by K-mean algorithm. The low frequency band of the image is segmented into three regions： sub-important region and background region. According to the feature of region, the region＇s degree of membership is determined, and then to achieve the multi-resolution representation of fusion result. The final image fusion result is obtained by performing the inverse multi-resolution transform. Experimental results indicate that the proposed method outperforms conventional image fusion method.