I want to detect the the object here below that is similar to a square but with rounded edges?
I can use morphological transform Gradient to find the contour with the largest area but some times even the other objects in the picture might be as BIG as the square object with rounded edges
Does anyone else have any other solution?
I played around with a few more colorspaces in OpenCV
LAB and YUV gave a nice shape of the square object(the square object was actually red in color)
Here's the result, both the LAB and YUV results were same.
Performed pixel wise operation using numpy.where in axis=2 and remove all the color pixels outside the interested square shaped object to [0,0,0] or black
This is the final result.
Hope it helps someone
O_O ow wait :D so you start with a question on shapes, then move yourself to color segmentation. We could have told you that from the start if we had the original image ^_^ nevertheless, good you solved it!
I would go for shape analysis in the following way
Hi, you can try to calculate the thinnes ratio: T = 4*Pi ( Area/Perimeter^2)
this measure has a maximum of 1, which corresponds to a circle, otherwise is more an rectangle. You can also find some other contour properties in OpenCV docu
Asked: 2018-03-23 02:35:56 -0600
Seen: 979 times
Last updated: Mar 23 '18
