A machine vision method for inspecting excess glue in spindle motor assembly

Abstract

In this paper, we propose a method to automatically detect excess glue in inner threads of bearing sleeve
of harddisk-drive spindle motor by machine vision techniques. The glue is used to join the bearing sleeve to
the cap of the motor. After the joining, the assembly then undergoes a hardening process. Excess glue from the
process may be present and will reside rigidly in the inner threads of the bearing sleeve. This prevents the
motor from securely fit into the harddisk-drive body resulting in scraping the workpiece. Oblique lighting is
employed to form reflected light back from surfaces of the excess glue. The defects generate highlights along
the threads in the image of the inner sleeve. To reduce spurious reflections from thread roofs, threads are
located by a geometric primitive fitting. By properly detecting those highlight spots within regions between
adjacent thread roofs, the defect is accurately identified. Experimental result showed a good performance of
our proposed method.