Rapid location of convex objects in digital images

This paper studies sampling strategies for the rapid location of objects in digital images, and shows how point sampling can be used to minimise computational effort. The process can be extremely efficient, especially when the image space is sparsely populated and large convex objects are being detected. In the case of ellipses, exact location is considerably aided by the new ‘triple bisection’ algorithm. The approach has been applied successfully to the location of well separated nearly elliptical cereal grains which are to be scrutinised for damage and varietal purity.