多分辨率分析耦合近似稀疏表示的图像融合算法

发布时间：2018-08-05 12:24

【摘要】：为了更好地处理图像高维特征奇异性,并兼顾融合图像目标特征与平均强度信息,提出了一种多分辨率分析与近似稀疏表示的图像融合算法。首先,对源图像进行对尺度分析,分别得到图像的高频和低频信息;然后,设计了近似稀疏表示(sparse representation,SR),通过近似SR系数来表示图像高频信息和低频信息;并利用绝对最大选择技术对近似SR稀疏转换,得到低频子带的近似系数和高频子带的细节系数,以达到用最少的系数来逼近奇异曲线。其次,构建了决策映射,对相同子带上的各SR系数的活性度和匹配度进行决策分析,输出决策值,通过决策值对图像进行匹配融合。最后,通过多尺度逆变换得到最终的融合图像。仿真实验表明:与当前图像融合算法相比,获得的融合图像具有更好的视觉效果,能有效图像突出目标信息,得到的图像具有更高的平均梯度和边缘评价因子;既突出了目标特征又保留平均强度信息,同时降低噪声影响。
[Abstract]:In order to better deal with the singularity of high-dimensional features of images and take into account the information of target features and average intensity of fused images, an image fusion algorithm based on multi-resolution analysis and approximate sparse representation is proposed. Firstly, the source image is analyzed to obtain the high frequency and low frequency information, and then the sparse representation SR is designed to represent the high frequency and low frequency information of the image by approximate SR coefficients. The approximate coefficients of the low frequency subbands and the detail coefficients of the high frequency subbands are obtained by using the absolute maximum selection technique to convert the approximate SR sparsity to approximate the singular curves with the least coefficients. Secondly, the decision mapping is constructed, the activity and matching degree of SR coefficients in the same subband are analyzed, the decision value is outputted, and the image is matched and fused by the decision value. Finally, the final fusion image is obtained by multi-scale inverse transform. Simulation results show that: compared with the current image fusion algorithm, the fused image has better visual effect, can effectively highlight the target information, the image has a higher average gradient and edge evaluation factor; It not only highlights the target characteristics but also preserves the average intensity information, and reduces the noise effect at the same time.
【作者单位】：广东工业大学计算机学院;中山大学信息科学与技术学院;广东暨通信息发展有限公司;
【基金】：国家自然科学基金(61272498) 广东省科技厅2014年度省前沿与关键技术创新专项(2014B010117002) 广东省中国科学院全面战略合作专项(2013B091500060)资助
【分类号】：TP391.41

【相似文献】