一种基于空频域特征的多传感器图像融合方法

提出了一种基于多分辨率小波分析和高斯 -马尔可夫随机场理论的边缘图像融合方法。在图像的多分辨率小波分析的基础上 ,基于高斯 -马尔可失随机场理论 ,在不同图像的对应区域 ,用回归分析的方法分别提取一组统计参数 ,这些参量表征了图像的局部结构特征 ,计算其相似性测度 ,最后由输入图像及其特征的相似性矩阵生成融合后的边缘图像。实验证明 ,这种融合方法具有较强的适应性和可靠性 ,即使在图像信噪比较低的情况下 ,也能取得较好的融合效果。

An Approach Based on the Features of Space-Frequence Domain of Image for Fusion of Edgemaps Obtained Through Multi-Sensors

In this paper, an approach based on multiresolution wavelet analysis and GMRF theory for the fusion of edge images obtained through multisensors is proposed. The hierarchical multiresolution wavelet information in conjunction with the context information extracted from GMRF constitutes local spatial features of images to be fused, then the similarities of these features are computed, the fusion result is achieved via computing a fusion function with respect to the similarity along with imput edge images. The effectiveness and reliability have been tested through simulations, especially on those images with low SNR.