Bogazici University Pattern Analysis and Machine Vision Laboratory |
|
|
|
The aim of this project is to identify the types of the tires that are transported on a conveyor belt using machine vision techniques. The photographs of the tires may be taken while they are moving or after they are stored in a black box. The photographs will be processed and the tires will be separated according to their types. There are 4 types of tires that are reported to be manufactured in the factory. The areas of the patches on the tires are good candidates for distinguishing between different types of tires. Thus, a labeling algorithm is applied on the patches and the areas of patches were counted in terms of pixel squared. A neural network is trained with obtained results. The neural network was able to separate the tires with 100 % confidence at the expense of long training time. However, the training of the neural network is carried out once, before implementing on the conveyor belt, thus extended training time is not effective on the run-time performance of the system. Some sample results are illustrated below. As can be seen, the labeling algorithm is appropriate also for the case of rotated tires. The patches in the vicinity of the center lines are considered in the classification procedure since the ones close to the edges are not labeled properly. For further information about the utilized labeling algorithm and neural networks, you can contact BUPAM Type 1:
Type 2:
Type 3:
Type 4:
|
Please send mail to dogozgur@boun.edu.tr with questions or comments .
|
