Coherent Motion Estimation for Stereoscopic Video

A method to estimate motion of pixels in an image pair with a computer, the method comprising:

• providing, by a computer, a first and a second video frame at a first time value, the first and the second video frames forming a stereoscopic image at the first time value;
• generating, by a computer and for at least one pixel, a first disparity value for the at least one pixel between the first and second video frames;
• storing the first disparity value in a first memory location;
• generating two motion vector parameters for the at least one pixel between the first and second video frames and a third and a fourth video frame at a second time value, the third and the fourth video frames forming a stereoscopic image at the second time value;
• storing the two motion vector parameters in a second memory location;
• generating a second disparity value for the at least one pixel between the third and fourth video frames at the second time value;
• storing the second disparity value in a third memory location;
• estimating a coherent motion of the at least one pixel, based on the first disparity value, the two motion vector parameters, and the second disparity value;
• and
• storing estimated coherent motion values in a fourth memory location,
• wherein the estimating a coherent motion of the at least one pixel comprises minimizing a cost function.