Circle Detection
As in the case of straight lines, circle features occur very widely in digital images taken from manufactured parts (including many food products). Again they can conveniently be located by using the Hough transform approach. This chapter describes the processes and principles that are involved and shows the means by which speed and accuracy can be improved.
• the basic Hough transform technique for locating circular objects in images.
• how the method has to be adapted if the radius is unknown.
• how accuracy of the center location can be improved.
• how speed of processing can be improved.
• how speed can be improved further by a line sampling procedure.
• how robustness only falls gradually as available signal decreases.
This chapter shows how the Hough transform is again able to provide a useful means for detecting a selected type of macroscopic feature in digital images. Again the method is seen to rely on the accumulation of votes in a parameter space, and the robustness of the technique to result from concentration on positive evidence for the objects.