Lecture: Thresholding and Binarizing Images
Slides
Image Thresholding
Open the class introduction slides in a separate window: https://stats4neuro.netlify.com/slides/03-image_thresholding#1
Remember, you can hit p to open up the preview mode and see the slides.
Post-Class
Please fill out the following survey and we will discuss the results during the next lecture. All responses will be anonymous.
- Clearest Point: What was the most clear part of the lecture?
- Muddiest Point: What was the most unclear part of the lecture to you?
- Anything Else: Is there something you’d like me to know?
Muddiest Points
how Otsu’s filter actually works. And examples of filter shapes at the end making different images would have helped clarify that - it didn’t make a ton of intuitive sense how that would work
Here’s a page that describes exactly how Otsu’s method is calculated:
http://www.labbookpages.co.uk/software/imgProc/otsuThreshold.html
We will cover filter shapes in lab.
still a little bit unclear on when neighborhoods overlap vs are non-overlapping. The way I’m thinking about it is that a neighborhood by definition is non-overlapping in normal thresholding, but adaptive thresholding uses overlapping neighborhoods? I think i’m going in circles and confusing myself more.
Normal thresholding doesn’t use neighborhoods at all - it’s an absolute threshold that is applied to all of the pixels.
Here’s another explanation for adaptive thresholding that hopefully will fill in the gaps: http://homepages.inf.ed.ac.uk/rbf/HIPR2/adpthrsh.htm
The math behind Otsu’s and Dilation and Erosion
On the idea of adaptive threshold: is each pixel in the image going to have a different threshold? For example on slide 14 of the lecture slides, the 4, highlighted in green, in the teal box will have a different threshold than the 4 to the left because that pixel will have different neighboring pixel values.
I’m still a little confused about the calculations behind erosion and dilation. (e.g. examples shown in slide 18 and 20)