基于洛伦兹范数的医学图像超分辨率重建研究

在超分辨率图像重建(SR)模型中,为了达到良好的重建效果,选择一个合适的代价函数是研究的重点。采用SR重建模型中的差错项选择了洛伦兹范数,正则化项选择了吉洪诺夫正则化,重建过程采用了迭代方法。提出的算法可以有效地解决医学图像SR重建过程中的去异值点和图像边缘保持的两大关键问题,达到良好的重建效果。为了验证上述算法的有效性,就一系列添加了运动模糊和不同噪声的低分辨率MRI医学图像进行了SR重建,并且与基于L2范数的重建算法的重建效果进行了比较分析。实验结果显示,所提算法具有良好的实用性和有效性。

Lorentzian Norm Based Super-resolution Reconstruction of Medical Image

In the Super-resolution reconstruction(SR) model, it is the key point of the research to choose a proper cost function to achieve good reconstruction effect. In this paper, based on a lot of research, Lorentzian norm was employed as the error term, Tikhonov regularization was employed as the regularization term in the reconstruction model, and iteration method was employed in the process of SR. In this way, the outliers and image edge preserving problems in SR reconstruction process can be effectively solved and a good reconstruction effect can be achieved. A low resolution MRI brain image sequence with motion blur and several noises were used to test the SR reconstruction algorithm in this paper and the reconstruction results of SRR reconstruction algorithm based on L2 norm were also be used for comparison and analysis. Results from experiments show that the SR algorithm in this paper has better practicability and effectiveness.