T. D’Orazio, M. Leo, M. Nitti, and G. Cicirelli (Italy)
Circle Detection, Neural Classifier, Real Time processing.
A large number of methods for circle detection have been studied in the last years for numerous image processing ap plications. The application domain considered in this paper is the soccer game. In these sequences of images it is very important to identify the ball in order to evaluate the goal event. This domain is challenging as a great number of problems have to be managed, such as occlusions, shad ows, objects similar to the ball, real time processing. In this work we have developed a visual framework that tries to solve the above problems mainly considering real time computational aspects. For this reason the ball detection algorithm has to be very simple in terms of time process ing but also efficient in terms of false positive rate. Our framework consists of two sequential steps for solving the ball recognition problem: the first step uses a modified ver sion of the directional Circle Hough Transform to detect the region of the image that is the best candidate to con tain the ball; in the second step a neural classifier is ap plied on the selected region to confirm if the ball has been properly deteted or a false positive has been found. Some expedients like background subtraction and ball tracking have been considered in order to mantain the search of the ball only in limited areas of the image. A large number of experiment has been carried showing that the proposed method obtains an high detection score.
Important Links:
Go Back
