What is subpixel edge detection ??

Hello

I am doing my seminar on subpixel edge detection.So i want some basic idea of subpixel and what are the step in matlab to do subpixel edge detection?

Answer 1

[Image Analyst]

It means trying to locate something to a pixel location that is not on a pixel, but between pixels. For example if you know you have a triangle, or Gaussian, or some other symmetrical peak, and your data is [10 20 30 40 50 60 60 50 40 30 20 10] you might estimate that the peak, with a value of 65, should be located at "pixel" 6.5 - between the two pixels that have values of 60.

Answer 2

[Yu Gao]

luckily, I'm working on the same issue for subpixel edge detection. As far as I found, there are two main streams of solution for this topic. First is Monent based method and the other is reconstructive approach. Monent based have spacial moment based method, Zernik monent based etc, Reconstruction basically fit a smooth function to the image gradient information. Firstly it needs accurate coarse edge then enhance it to subpixel level. Coarse edge is more or less similar to the classical sobel, gaussian or LoG method. The second step splited as I mentioned before Monent based or fitting based. Moment based can be assumed as conv image with a moment kernel and some of literatures illustrate different moment based method have similar results. I didn't touch fitting method yet. I hope this helps.