Hello, guys!
Now I am working on a task to detect and tracking many tennis balls in the ground.
The recognition effect is very good by using cvblob. But there is one problem, when
several balls overlap, it can only found one blob, I am considering how to solve this situation.
My idea is to continue applying the randomized hough transformation for the obtained polygon.
As the general hough transformation requires the input as Iplimage*, and the polygon is vector<cvPoint >
style, could anyone tell me how to transform a vector<cvpoint> to Iplimage*.
As we see, from the above picture, when two tennis balls overlap, it also show one blob.
If possible, could anyone give me some good suggestions on how to solve this situation?
Thank you for all in advance! Have a nice day !!!