Difference between revisions of "Projects:2020s2-7233 Image Denoising with Dictionaries"
| (6 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
== Abstract == | == Abstract == | ||
| − | + | Image denoising is essential phase in the pre-processing of images. Different algorithms have been proposed in the past few decades with varying denoising capabilities. K-Singular Value Decomposition (K-SVD) algorithm is currently the most widely used adaptive learning dictionary in image denosing problem. In this study, K-SVD algorithm will be implemented in image denoising and observe the performance when using different image datasets. | |
== Introduction == | == Introduction == | ||
| − | + | Dictionary learning is a topical area of signal processing research. It shares similarities but also crucial differences from other machine learning ideas such as deep learning neural networks. It rests on the idea of sparse representations using a well-designed dictionary that can lead to high performance in a wide range of signal processing applications. Image denoising is one of many signal processing applications. The idea developed here can easily be transferred to other application area. | |
| − | == Project Team == | + | === Project Team === |
| − | === Project Students === | + | ==== Project Students ==== |
* Muhammad Haziq Saharuddin | * Muhammad Haziq Saharuddin | ||
* Muhammad Faizal Azhar | * Muhammad Faizal Azhar | ||
* Chen Chen | * Chen Chen | ||
| − | === Project Supervisor === | + | ==== Project Supervisor ==== |
| − | == Objectives == | + | * Associate Professor Brian Ng |
| + | === Objectives === | ||
| + | * To analyse performance of dictionary learning on various image datasets | ||
| + | * To implement K-SVD method image denoising | ||
| + | * To evaluate the effectiveness of different matching pursuit of algorithms in dictionary learning | ||
Latest revision as of 23:29, 14 September 2020
Contents
Abstract
Image denoising is essential phase in the pre-processing of images. Different algorithms have been proposed in the past few decades with varying denoising capabilities. K-Singular Value Decomposition (K-SVD) algorithm is currently the most widely used adaptive learning dictionary in image denosing problem. In this study, K-SVD algorithm will be implemented in image denoising and observe the performance when using different image datasets.
Introduction
Dictionary learning is a topical area of signal processing research. It shares similarities but also crucial differences from other machine learning ideas such as deep learning neural networks. It rests on the idea of sparse representations using a well-designed dictionary that can lead to high performance in a wide range of signal processing applications. Image denoising is one of many signal processing applications. The idea developed here can easily be transferred to other application area.
Project Team
Project Students
- Muhammad Haziq Saharuddin
- Muhammad Faizal Azhar
- Chen Chen
Project Supervisor
- Associate Professor Brian Ng
Objectives
- To analyse performance of dictionary learning on various image datasets
- To implement K-SVD method image denoising
- To evaluate the effectiveness of different matching pursuit of algorithms in dictionary learning
