I am working on a program which aims to detect the crack on bearings.I have tried some of the relevant methods in Vision Assiatant,but it can detect not only the cracks,but also edges,shade and some other elements.So I want help,have you made such a detection which are made to detect cracks and can locate the crack only?That is to say,your procedure can automatically locate crack and detect crack only.And could you please send the procedure to me?Thank you very much.
You can get this by thresholding. But the lighting you are using is not correct for the application. Use a low angle lighting ar try a dome light. Please make sure the lighting is as uniform as possible.
First of all,thanks for your attention and your suggestion.And can you describe your suggestion in more detail?Thresholding is used to get binary picture,but the crack' grayscale is probably in tne middle level,therefore we may highlight irrelevant details but neglect cracks.I hope you can provide some guidance.Thank you very much.
I have tried texture detection to extract the crack,and my Vision Assistant script is attached.But some edges and particles can also be highlighted.I just want to detect the crack.Do you have any advice?I need your help.
There is a 3-part series which introduces the machine vision lighting, and in the third part of this series, illumination techniques are explained which I think can help you with your application: http://www.ni.com/white-paper/6903/en/
As mentioned in the above suggestion, low angle/dark field lighting can be used to detect surface defects. Once you have prepared the suitable lighting, you can easily detect the cracks that you have shared. I can suggest you to try using edge detection functions in order to do that.
