Abstract
Automatic defects inspection and classification in ceramic tile plays a crucial role in ceramic tiles industry. In order to improve the required computing time to detect and classify spot and crack defects in ceramic tiles, this paper proposed parallel algorithms based on the graphical processing unit. The proposed algorithm divides ceramic tile images into non-overlapped partitions, identifies the defected partitions and then carries out the defect classification algorithm only on the defected partitions. Simulation results indicate significant improvements in terms of classification time in comparison to the current techniques. Furthermore, computation time for parallel algorithm excels the sequential algorithm. The results show the speed up of the computation of the proposed parallel compared with sentential algorithms that increase as the image size increases. Finally, the speed up of 45 times has been accomplished.