ICA APPROACH TO IMAGE AND SIGNAL PROCESSING

Student: Kong Chiwai

Abstract

The purpose of our software package addresses the image processing. It will read any formats of image files and use ICA, ISA or TICA as the principal model for natural images. Finally, we will see that these algorithms do provide some models that are very similar to the most sophisticated low-level image representation used in image denoising. 

The implementation of ICA requires selecting suitable parametrizations and estimation of parameters. Testing all possible combination of parameters would require a huge amount of experiments. In order to restrict their number we test the influence of one or two parameters at a time. We also propose a graphical user interface (GUI), which greatly reduces the amount of the test time.

Project Goals

 The following problems will be solved for the duration of this project:

• Design an algorithm, which can restore the original image from the noise-free image of the result. 

• Improving performance of ICA model such as a speed of iteration and the consumption of memory.
• Add one more algorithm that is based on Principal Component Analysis (PCA).
• Add the filer functions.
• Add plotting 2D and 3D graph functions.
• Add error calculation.
• Extend the current GUI in order to accommodate the above functions.
• Incorporate the results into the CSC 470 Introduction to Computer Graphics.