Abstract :
The first phase of a feasibility study of automatic assembly and inspection of incandescent light bulb filaments in the industry is described. The problem is limited to the visual aspect of the system. Picture-processing and pattern-recognition techniques are employed 1) to preprocess the signal provided by a television camera, 2) to transform and extract the features of the object, 3) to classify the lead wires and support wires, and 4) to specify the coordinates of the locations of the tips of the wires. On the assumption that the locations of the tips in the three-dimensional space can be determined through the analysis and processing of two or more two-dimensional visual picture patterns at different angles, the first phase of this feasibility study is confined to processing two-dimensional picture patterns. A complete set of algorithms has been developed, implemented, and tested in the Signal Processing Laboratory, Case Western Reserve University, Cleveland, Ohio. Repeatability and consistency of the system are investigated by varying some key parameters, such as lighting, orientation of the object, and setting of the camera. The experimental results show no error in classification and negligible error in specification of the coordinates of the tips.
