Drowsiness Detection Using Dynamic Eye Thresholds

The patent applies dynamics by transforming the static fixed threshold into a dynamic adaptive threshold that automatically adjusts according to the detected eye size in captured images. The system calculates the eye size from each captured image and dynamically updates the threshold accordingly, enabling the detection system to adapt to different drivers and driving positions without manual intervention.

The system implements self-service by automatically calibrating the detection threshold based on the actual eye size detected in the captured images. The apparatus performs self-adjustment through the threshold calculation unit, which computes the appropriate threshold from the image data itself, eliminating the need for external calibration or manual threshold setting by operators.

The patent applies preliminary action by performing threshold calibration during the initial phase using captured facial images. The system establishes the baseline eye size and detection threshold before actual drowsiness monitoring begins, allowing subsequent detection to proceed more efficiently with pre-established reference values that reduce real-time processing requirements.

The system uses partial action by selectively processing only the eye region from captured facial images rather than analyzing the entire image. The eye region extraction unit isolates and processes only the relevant eye area, reducing computational load while maintaining detection accuracy by focusing resources on the critical detection zone.