Zhang Rui, Wang Zhongyang, Sun Haoze, Deng Lizhen, Zhu Hu
Jiangsu Province Key Lab on Image Processing and Image Communication, Nanjing University of Posts and Telecommunications, Nanjing 210003, China.
National Engineering Research Center of Communication and Network Technology, Nanjing University of Posts and Telecommunications, Nanjing 210003, China.
Sensors (Basel). 2023 Jul 23;23(14):6616. doi: 10.3390/s23146616.
In this paper, a unified optimization model for medical image fusion based on tensor decomposition and the non-subsampled shearlet transform (NSST) is proposed. The model is based on the NSST method and the tensor decomposition method to fuse the high-frequency (HF) and low-frequency (LF) parts of two source images to obtain a mixed-frequency fused image. In general, we integrate low-frequency and high-frequency information from the perspective of tensor decomposition (TD) fusion. Due to the structural differences between the high-frequency and low-frequency representations, potential information loss may occur in the fused images. To address this issue, we introduce a joint static and dynamic guidance (JSDG) technique to complement the HF/LF information. To improve the result of the fused images, we combine the alternating direction method of multipliers (ADMM) algorithm with the gradient descent method for parameter optimization. Finally, the fused images are reconstructed by applying the inverse NSST to the fused high-frequency and low-frequency bands. Extensive experiments confirm the superiority of our proposed TDFusion over other comparison methods.
本文提出了一种基于张量分解和非下采样剪切波变换（NSST）的医学图像融合统一优化模型。该模型基于NSST方法和张量分解方法，融合两幅源图像的高频（HF）和低频（LF）部分，以获得混合频率融合图像。一般来说，我们从张量分解（TD）融合的角度整合低频和高频信息。由于高频和低频表示之间的结构差异，融合图像中可能会出现潜在的信息损失。为了解决这个问题，我们引入了联合静态和动态引导（JSDG）技术来补充高频/低频信息。为了提高融合图像的结果，我们将乘子交替方向法（ADMM）算法与梯度下降法相结合进行参数优化。最后，通过对融合后的高频和低频波段应用逆NSST来重建融合图像。大量实验证实了我们提出的TDFusion相对于其他比较方法的优越性。
