In this paper, we present an advanced machine vision system for VFD inspection.  The system includes two procedures—off-line learning and on-line testing.  The learning procedure can be either automatic or manual to give user more flexibility to control.  A template is generated from the intelligent learning, which includes all features needed by the VFD inspection.  The automatic testing procedure loads the template and captures the images of the VFD boards on the manufacturing line, then gives the accurate results quickly.  Some machine vision algorithms such as image binarization, image tilt detection, and component detection are described in the paper.