I want to write a program that can identify 4 different shapes when given an input. My plan is to program an FFT algorithm so I can use the frequency contents of the shapes to classify them using k-means. I then want to use Python Turtle to come up with datasets of shapes. (Unless you guys know of available datasets). I was going to use k-nearest neighbors for the main program that would assign new inputs to the proper cluster and output what shape it is.
I was hoping for some advice on how to ensure I get a working program given my above plan and whether or not this sounds feasible.
Some issues I am thinking of: 1. Aliasing with the FFT 2. What image format should I use? Ideally I think it should be something simple like a bitmap. 3. Are there better ways to come up with datasets? 4. Normalization issues?
Also I am required to use C for this project.