Abstarct: In this research, granulated curve of aggregates is derived by making use of image processing and neural network techniques when they are falling from a conveyer belt. To achieve this, the film, captured with high frxame rate (240 frxames per second) is converted into several frxames and individual images are selected for further process. Boundaries of aggregates in each individual picture are distinguished by utilizing the Support Vector Machine (SVM) technique. Afterwards, size of aggregates is calculated and granulated curve is carried out. The results are compared with the experimental tests and the discrepancies are illustrated and discussed.