Aggregate Classification by Using 3D Image Analysis Technique
Aggregate occupy approximately 80 percent of the total volume of concrete mix, and aggregate physical characteristics significantly affect the properties of concrete both fresh and hardened state. Selection of improper aggregates such as flat and elongated particles may cause failure or deterioration of a concrete structure. Therefore, selection process of aggregates for a specific job is very important. There is no standard test method for evaluating the aggregate physical properties effectively. The manual standard test methods (EN 933, ASTM D 4791, ASTM C 1252, and ASTM D 3398) are laborious, time consuming and tedious measurements. Trent to tighten specifications for aggregate properties along with recent technological advances in technology, availability of high performance computers, and low cost imaging systems support usage of image analysis methods for quantitative measurement of aggregate properties such as size, shape and texture with easy, fast, real-time and without human errors. In last decades, two dimensional (2D) and three dimensional (3D) image analysis techniques have been used to measure size, shape, and texture of aggregates. In this paper, shape and size parameters (features) of four different types of aggregates are calculated by 3D image analysis technique and aggregates are classified by three different artificial neural network models with using these parameters. Best classification performance is given by a multilayer perceptron method which is 90,84 % precision.
Keywords: Aggregate, Shape, Image analysis, 3D, Classification