Introduction to Blink Detection and Tracking of Eyes For Eye Localization Seminar Topic:
We need eye normalization so that we can read anything properly, we can log into computer that are based on the iris detection in mainly security systems identifications. For this certain steps are followed, first thresh holding process, then frame differences process, then eye is localized and then eye are tracked.
In thresh holding face image is captured then b/w photo is captured on mainly focused on the eye. Then other part accept eye is rejected and main focus of eye is taken then their small parts on the graph are plotted.
Algorithms: first we take the possible motion using the frame differencing technique, and then thresh holding the motion regions for morphological operations. Remove blobs that are unsuitable in width or too small. Mark correct direction of motion of blob. Reject the blob that is not suited near the location of the blobs with downward motion. They would represent eye during closure of iris. If the motion is downward then steps are followed again.
If the motion was detected upward then classify the frame upward where the motion was detected downwards to the frame as blink frames. Process is started after new frame.
Eye tracking: in eye tracking they have 20 features point which gave us more accurate result. These featuresare taken from eye and they are tracked in the different places for reinstallation. First they initialize eye region and they track it before eye blinks. For speeding up localization process speed has to increase the eye tracking and optical flow.
Conclusion: in this way we have studied the eye localization technique by using both magnitude and direction of flow vectors. This is accurate and fast method for locating and tracking.
Download Blink Detection and Tracking of Eyes For Eye Localization PPT .