Enhanced video-based driver monitoring using phase detect sensors

A method of remotely monitoring a head pose or eye gaze of a subject of a vehicle using a single camera, the method comprising:

• a) receiving, from the camera, a plurality of time-sequenced images of the subject's face;
• b) determining a two dimensional position of the subject's eyes in each of the images;
• c) determining, for at least some of the images, a spatial image offset by comparing first and second image data generated by a plurality of phase detecting pixels in the digital image sensor;
• d) determining a distance between the subject's face and a digital image sensor of the camera based on the spatial image offset;
• e) determining for at least some of the images, based on steps b), c) and d), a three-dimensional position of the subject's face or eyes;
• and
• f) determining a head orientation or eye gaze of the subject based on the three-dimensional position of the subject's face or eyes.

A system configured to measure a distance of a face of a vehicle subject in a subject monitoring system, the system comprising:

• a single camera, having an imaging system, positioned to capture an image of the subject including the subject's face, the camera including a digital image sensor having pixels including a plurality of phase detecting pixels, the phase detecting pixels configured to generate first and second image data corresponding to light received along two optical paths through the camera's imaging system;
• and
• a processor circuit configured to perform operations comprising:
• controlling the camera to capture an image of the subject including the subject's face;
• processing the image to identify a face region of the image as being a portion of the image that was generated by a region of pixels corresponding to the subject's face or head;
• determining, within the region of pixels, a first subset of the phase detecting pixels as those phase detecting pixels which correspond to the face region;
• comparing first and second image data generated by the first subset of the phase detecting pixels to determine a spatial image offset;
• and
• determining, from the spatial image offset, a first distance estimate of a distance between a region of the subject's face and the digital image sensor.