Image Processing

1. Image processing is often used on raw data from many kinds of microscopy. Fourier transforms are of great use in this context Links to an external site..

The image hw8/denoise/sample_image.png is of a bunch of wavy tubes that has been blurred and has noise added. (For example this might be of DNA chains on a substrate.) The job you have is to process the image well enough so that you can distinguish individual tubes from each other. The script in the same directory /hw8/denoise, has most of the code already there. You have to figure out how to filter the image (in k space) so that you can distinguish the individual tubes.

2. There are practical problems associated with image analysis that should be taken into consideration.

• Which techniques or biological systems are these problems relevant to?
• To process this image, you had to apply a filter to it so that you could see individual lines. Discuss potential problems with applying a filter, in particular, ringing artifacts Links to an external site.. Could you see such effects producing spurious results in biological images. If so, how?
• An image can be under-sampled creating aliasing. Explain how this artifact comes about and how it is related to Moiré patterns.
• For both of the above artifacts, explain how you might try to distinguish if a feature you were observing from an experimental image was real, or a result of such problems.
• Another problem with STM images on surfaces are artifacts associated with the substrate, for example see here Links to an external site.. Discuss how this can influence the resultant images. Did such artifacts lead researchers astray?