摘要
为了克服传统方法去噪会损失部分有用信息的缺点,该文采用稀疏编码和字典训练两个关键技术,准确又高效地区分开图像的有用信号和噪声信号,更好地实现了去噪。针对字典训练的过程利用了K-SVD算法,研究了其原理和去噪流程,由于字典学习是通过机器学习获得而不是预先选定得到的,从而可以更完整地保留图像原有的信息,最终获取更高的峰值信噪比。通过对不同算法的仿真分析,验证了该方法的有效性。
In order to overcome the disadvantages of the traditional methods denoising which lead to lose some of effective information,sparse coding and dictionary training are used as two key technologies. This method can distinguish the effective signal and noise signal of the image accurately and efficiently, and better realize the denoising.In view of the K-SVD algorithm that is used in dictionary training, the principle and denoising process are researched.Since dictionary learning is obtained by machine learning rather than pre-selected, it can not only better reserve the original information of the image, but also achieve higher Peak Signal to Noise Ratioultimately.The effectiveness of the method is verified by simulation analysis of different algorithms.
出处
《电脑知识与技术》
2018年第1Z期164-165,共2页
Computer Knowledge and Technology
关键词
机器学习
字典学习
稀疏表示
K-SVD
图像去噪
machine learning
dictionary learning
sparse representation
K-SVD
image denoising
