Title: Non-linear Noise reduction
Name: Junjie Wang, HaiXin Tie

Linear noise reduction algorithms are best for filtering additive 
white Gaussian noise (AWGN). However, it causes loss of detailed
information (high-frequency component) in the image, which in some
cases is undesirable. Motivated by this property, nonlinear filtering
has been widely investigated as a noise reduction technique. Another
desired property lies in the flexibility of this technique. However,
it is also well-known that nonlinear filtering doesn't work well with
the AWGN case. 

In this project, we will investigate the application of nonlinear noise
reduction techniques presented in literatures. Most of the approaches 
in this area can be categorized into 3 classes: spatial domain, frequency
domain, and information-theoretic approach [1,2,3]. We will implement 
some algorithms proposed in the literatures, and try to improve the
performance by integrating the different approaches. We will first 
focus on the AWGN case to evaluate our methods, but some other noise
models will also be considered.


Reference

1. R. C. Gonzales and R. E. Woods, Digital image processing, 
   New York, Addison-Wesley, 1992.

2. A. Beghadai and A. Khellaf, ``A Noise-filtering method using
   a local information measure'', IEEE Trans. Image Proc., vol.6,
   no. 6, 1997.

3. J. Lin, N. Ansari, and J. Li, ``Nonlinear filtering by threshold
   decomposition'', IEEE Trans. Image Proc., vol. 8, no. 7, 1999.