The random cone winding system is one of cmmonly used to rewind the produced yarn from the spindel to the cone. In the random system of cone winding, the patterning status leads to many problems in further process, such as yarn breakage or early take off of the package. In this research, computer vision has been applied as an online system to detect patterning status by using power angle spectrum of fuorier. To survey capability of the computer vision system, various samples of cone packages were experienced. The results presented that the power angle spectrum of sample images moved to the maximum point where, the patterning status would be occurred. The system was adjusted to stop cone winding process while the value of power angle is increased more than 0.25. The adjusted system could detect patterning status before occorance regarding to all of experienced samples correctly.
