Isa، نويسنده , , Nor Ashidi Mat and Sani، نويسنده , , Zamani Md and Al-Batah، نويسنده , , Mohammad Subhi and Mat Isa، نويسنده ,

3268005

Automated Intelligent real-time system for aggregate classification

10010

Traditionally, mechanical sieving and manual gauging are used to determine the quality of the aggregates. In order to obtain aggregates with better characteristics, it must pass a series of mechanical, chemical and physical tests which are often performed manually, and are slow, highly subjective and laborious. This research focuses on developing an intelligent real-time classification system called NeuralAgg which consists of 3 major subsystems namely the real-time machine vision, the intelligent classification and the database system. The image capturing system can send high quality images of moving aggregates to the image processing subsystem, and then to the intelligent system for shape classification using artificial neural network. Finally, the classification information is stored in the database system for data archive, which can be used for post analysis purposes. These 3 subsystems are integrated to work in real-time mode which takes an average of 1.23 s for a complete classification process. The system developed in this study has an accuracy of approximately 87% and has the potential to significantly reduce the processing and/or classification time and workload.

41

Aggregate classification , Intelligent system , real-time system , Artificial neural network , image processing , Shape classification

Studia Iranica

50

50