Enriched algorithmic approach for classification and recognition of part families for medium scale industries
Abstract
In this paper, an interdisciplinary approach is proposed. This collaborative approach is applied for classification and recognition of mechanical components. These mechanical components are based on part families used for designing and manufacturing in medium scale industries. The major objective of work is to recognize and classify the components such as bolts, nuts, and washers and to evaluate design and manufacturing attributes in real-time. The proposed real-time Industrial Component Detector (ICD) system is developed to process the input component using methodologies based on template matching, clustering techniques, and region extraction technique. Aforesaid methodologies are applied for designing and implementing three algorithms using C# & .NET namely; Comp-Temp-Match, K-Cluster-Comp-Match and biggest BLOB algorithm respectively. The experimental results in the system for recognition rate is achieved using aforementioned techniques is 97%. Moreover, in this work a system is proposed which outperforms using referencing model, since the model doesn't require to be calibrated again and again, and also performs auto calibration.
- Publication:
-
American Institute of Physics Conference Series
- Pub Date:
- July 2021
- DOI:
- 10.1063/5.0058048
- Bibcode:
- 2021AIPC.2358j0016S