Computer vision system for object tracking and time-to-collision

A method for vehicle perception for a sensor processing system, comprising:

• receiving image data from a camera, wherein the image data comprises a current image and a next image;
• extracting contours from the image data;
• determining a first contour of the current image along a plurality of image row boundaries, wherein the first contour comprises a plurality of first contour points linked to each other via respective first edges between each first contour point,
• determining a second contour of the next image along the plurality of image row boundaries, wherein the second contour comprises a plurality of second contour points linked to each other via respective second edges between each second contour point,
• applying an image velocity vector to project each of the plurality of first contour points to the next image, wherein at least some of the projected first contour points are not located on an image row boundary;
• processing the projected contour points to determine projected edges forming a projected link between each projected contour point;
• moving the projected contour points along the projected link to align each of the moved projected contour points to a respective image boundary;
• and
• linking each of the moved projected contour points to each of the second contour points.

A non-transitory computer-readable storage medium, including a set of executable instructions that causes a vehicle sensor processing system to:

• receive image data from a camera, wherein the image data comprises a current image and a next image;
• extract contours from the image data;
• determine a first contour of the current image along a plurality of image row boundaries, wherein the first contour comprises a plurality of first contour points linked to each other via respective first edges between each first contour point,
• determine a second contour of the next image along the plurality of image row boundaries, wherein the second contour comprises a plurality of second contour points linked to each other via respective second edges between each second contour point,
• apply an image velocity vector to project each of the plurality of first contour points to the next image, wherein at least some of the projected first contour points are not located on an image row boundary;
• process the projected contour points to determine projected edges forming a projected link between each projected contour point;
• move the projected contour points along the projected link to align each of the moved projected contour points to a respective image boundary;
• and
• link each of the moved projected contour points to each of the second contour points.