My Institute Technical Summer Project required Image Processing for detection and classification of coins on a carrom board. It uses the Hough Transform for detecting circles(and provides their radii too).
Thresholding is done first, and then followed by "Edge Detection" as in openCV. Using the obtained edges, a circle is detected(if it falls within the parameters of the function). The main(important) parameters are Minimum and Maximum circle Radii and the Minimum possible distance between two circles.
-
Auto-Calibration: User is asked to point to as many similar colored pieces as possible(For each:Red, Black and White coins). This allows me to calculate range of values of pixels corresponding to that particular colour. Adds flexibility of environment and lighting condition.(Note: HSV is used for classification/thresholding).
-
Camera Input: You can take input from the default camera of the computer. This enables you to attach a USB webcam for the purpose of image capture. It was added primarily due to its requirement in my specific application.