What are some easy methods to find objects in image processing?
I have the freedom to choose the type of object I want to detect, so I went with green circle targets, however no matter the pick, it becomes really hard to detect or even see when the sunlight shines directly on the target. Also, circle doesn't really work if I try to use any corner detection algorithms. Any ideas on what works best?
Actually using a colored circle is quite basic. Combined with a RGB camera, it will indeed fail with many lighting conditions. Why not try to find a target which is marked by a QR code? That will allows you to have a big contrast black on white for example, and by then performing histogram equilization you can counter the sunlight influences.
What about first detecting the circles and then with color values you can use detect green , as far as sun shine concern we can give a range of values to green color like if green color having value "1" then lets suppose with sun shine it become more light so cann't we suggest the program like if color goes light from 1 to 0.5 it that particular circle then its green as well .
As far circle detection concern : Study the curvature of every contour pixel and check if it fits a circle or ellipse. This check may be done by computing a histogram of edge orientations for the contour pixels, or by checking the gradients of orienations from contour pixel to contour pixel. In the second case, for a circle or ellipse, the gradients should be almost uniform . Refrence
For Different color detection