MNIF, Wael (2016) Dictionary learning to retore images PRE - Research Project, ENSTA.



There are many techniques to restore images. We can learn a dictionary to do so. K SV D is a method that help to restore image using a dictionary. In this internship, We choose to work with a sparse dictionary. The K-SVD is minimising l2 norm. So, we were trying to improve this method replacing l2 with l1. So, we get a method called l1-KSVD. The result we obtained using this method can be better than the result obtained with K-SV D but with a higher calculation time. The previous methods were used to restore images with gaussian noise. But, sometimes we face images with poisson noise. So, we have to find other function to minimize to restore image with poisson noise.

Item Type:Thesis (PRE - Research Project)
Uncontrolled Keywords:Dictionary learning, sparse, K − SV D, ` 1 − KSV D, gaussian noise, poisson noise
Subjects:Mathematics and Applications
ID Code:6744
Deposited By:Wael Mnif
Deposited On:13 oct. 2016 13:16
Dernière modification:13 oct. 2016 13:16

Repository Staff Only: item control page