理学院“格物论坛”学术报告（二十六）

理学院“格物论坛”学术报告（二十六）

报告题目: Patch Based Image Denoising Methods

报告人： 黄玉梅   教授     兰州大学

报告时间： 2021年5月22日（周六）10：00-11：00

报告地点： 理生楼A502

报告人简介：

黄玉梅，兰州大学数学与统计学院教授，博士生导师。2008年博士毕业于香港浸会大学。研究方向为图像恢复和数值代数，主持国家自然科学基金（青年）项目1项和面上项目2项，在图像处理和数值计算国际期刊发表论文30余篇，其中包括SIAM Journal on Imaging Sciences，IEEE transactions on Image Processing和SIAM Journal on Scientific Computing

报告内容简介:

Matrix low-rank minimization helps in exploiting the low-dimensional data from the high-dimensional data. The application of matrix low-rank minimization in image denoising is built upon nonlocal self-similarity based methods for image denoising. In this talk, we consider a rank minimization problem where the matrix rank is directly used as the regularization term in the cost function. The solutions of the rank minimization problem are analyzed and obtained by a hard-thresholding through singular value decomposition. Then by exploiting patch-based nonlocal self-similarity scheme, we apply the proposed low-rank minimization method to remove the noises in images. The experimental results illustrate that the proposed method can remove image noises very well and it is competitive to the existing state-of-the-art noise removal methods in the literature.

欢迎广大师生 ！

科学技术处、理学院数学系