| Automated visual inspection systems are an indispensable component of modern manufacturing processes. A machine visual system eliminates any errors related to human fatigue and accommodates increases in production without increasing the long term cost. In this thesis a machine-based visual inspection system for cardiac pacing and defibrillation leads was developed and tested. The main objective of this research is to develop the methods and algorithms necessary to inspect dimensions of cardiac pacing and defibrillation lead connectors. To create the inspection vision system it is necessary to select a proper illumination technique, camera, lenses and image processing algorithm. This thesis presents the design of such a vision system for the dimension inspection of the connector of a particular lead. In the algorithms, pattern recognition is used over filtered images to identify connector features. The connector dimensions are checked to corroborate that they comply with the ISO standard. To determine the error percentage we measure each connector with a caliper and compared those values with the ones given by the vision system. The error percentage values are between 0% and 13%. The error percentage is less than 1% in the majority of the connector's parts. These results were obtained using a lowpass filter in series with a Nth order filter over images with pixel resolution of 0.0808 mm/pixel. The developed machine vision system proved to be reliable with high speed, high quality, and high flexibility. |
