An image processing system for reading time from the optical image of hour plate

In this study, a time reading image processing system, conceiving time from the optical image of a watch with an analog indicator (hour plate) was implemented. By time reading system, the image of a watch with hour hand, minute hand and second hand is captured with a video camera that is positioned against the watch. The time is read by analyzing the image with the selected image processing techniques. The speed of analysis is acceptable for real time operation.The system, automatically reads the time from a watch with analog indicator. That process basically consists of finding the value pointed by analog indicator by sensing the positions of pointers which are inside the image of the analog watch. Various image processing techniques are used to recognize hour hand, minute hand and second hand that are the parts of watch indicator. Second hand is segmented and recognized from the image by motion based segmentation. The position of watch in the image frame is found by using features extracted from the skeleton of second hand. Hour hand and second hand are segmented from the image by seeded region growing technique. By using features extracted from both hour hand and minute hand and their skeletons, two hands are differentiated and recognized. Finally by using features extracted from the skeleton of hands, the time is conceived.Keywords: Image Processing, Object Recognition, Image Segmentation, Thining, Seeded Region Growing