-4

I have a sinuous signal in my image that you watch in below image. (in other words more than one signal with it's shadow)

My image in Matlab

I want to extract this sinuous signal (such a 1d signal) and find it's duty cycle, frequency and etc. As you can see there is a lot of noisy pattern also in my image and finding threshold or finding edges did not help me to solve the problem.

Also, I want to have a automatic algorithm that have generalize for my other images such this. Therefore,I think the finding of ROI couldn't be done, because in general sinuous signal begin from different row.

glennsl
  • 28,186
  • 12
  • 57
  • 75
Bashid
  • 433
  • 5
  • 15

1 Answers1

1

What you said, is quite a project, not a question. I'd rather to study more about the similar projects and ask about some details in here. Although I could give you some hints:

1-Do a suitable thresholding on your image, like Otsu method (with a little change in the threshold value if it's needed).

2-On the binary image do an erode and a dilate successively to get rid of the small objects.

3-Do a horizontal dilation to get all parts of a curve stuck together. This gives you two firm curves.

4-Find connected components of each curve and with their prositions, find the lowest pixels (local minimum of signals)

5-Horizontal position of these local minimums, gives you the frequency.

MeiH
  • 1,763
  • 11
  • 17