Some challenges we faced was deciding on how to choose the optimal filter parameters. It was important to balance between over-smoothing the image and effectively removing noise. We had to carefully determine filter sizes and the correct thresholding values for wavelet de-noising to ensure the best results.. Getting poisson noise as also a challenge, as a lot of the noise in Matlab did not change the image structure a lot, and would be an easy fix for the filters, which might have been caused as it affects more of the low-intensity parts of the image, and we could not tell which parts of the image had been modified.
We have split up the next weeks tasks. We need to work on developing the website, as well as mainly the wavelet denoising and bilateral filtering. These are both image processing techniques, and we also need to assess how they affect the image and test different methods to get the best possible result.
One thing that we have learned is how to create noise in any image using the 'imnoise' command in matlab. This command is important/relevant to our project because we need images with noise to be able to denoise the image.Â
We also plotted difference performances using different filters and de-noising techniques, to see which techniques would fit which sort of noise introduced.