Mariah Douglas
douglasm@leland.stanford.edu
EE 368 Project Proposal
I intend to implement two distinct non-linear noise reduction models.  I have 
read a few articles in IEEE Transactions on Image Processing concerning noise 
reduction, but I have not yet decided on two specific algorithms to implement.
Once I have decided on two algorithms, I will focus on applying these 
algorithms to three test images, namely lena.tif, man.tif, and elaine.tif, 
which I will first corrupt with three different levels of noise (I have not 
yet decided whether I will treat quantization noise or noise due to bit 
corruption.)  Note that, by selecting these test images, I am effectively 
limiting my scope to noise reduction in grayscale images. I have chosen 
these images because I believe them to be sufficiently complex and distinct 
to allow a more complete evaluation and comparison of the algorithms
performance than one based on a single image alone. As for improving the 
performance of an algorithm, I cannot yet comment on which algorithm I will 
improve or in which manner I will attempt to improve it, since I have not
yet selected an algorithm.