Differentiation of objects using camera vision can be done by which method? template matching etc

lakshmip001

I am going to work with Robotiq camera sensor. I was searching how to program a robot using a camera sensor to differentiate size of objects. When I went through I found that I can use this camera sensor for locating an object but didn't get how to differentiate between two objects using size of an object.
 
Please guide how I can use camera sensor to differentiate objects using size of the objects in the snapshot given.

Annick_Mottard

Hi @lakshmip001 !

You would need to teach the objects you want to detect to the camera. Your program will then have as many Camera Locate commands as there are parts you taught it. This way, the camera would look for each part, in the order they are placed in the program. If you use a programming logic, you can track which part has been detected and perform object sorting this way. 

Would this be suitable for your application?

Annick_Mottard

@lakshmip001 the smallest dimension of the part should indeed be at least 7.5 mm, not its area. You can refer to this DoF post for more details. 

lakshmip001

Hi Annick_Mottard

Thanks for the help. I can locate the object of different shapes by giving the reference image. If I am going to change shape I need to be changing the reference image. Can I know if I can do automated with the change of reference image for each shape. I hope you can understand it. Any help is appreciated.

Thanks,
Lakshmi

Annick_Mottard

Hi @lakshmip001 ,

You got this right : the Camera will do part and feature detection of the different parts you will have taught it. The rest of the operations depends on the programming logic you put behind it. 

From Polyscope, in a robot program, the Camera Locate node will act as an "if" statement. If the Camera detects the part or feature it has been taught, the program will perform the commands within this node. Thus, you could add a flag or a state variable that allows you to track which part has been detected. 

lakshmip001

Hi, Minimum Field of View for Robotiq camera is 10cm x 7.5 cm. Object should be 10% of Field of View to detect the object. So minimum area the object should occupy is 7.5 cm^2. Am I right?