Method of concurrent multiple-mode motion estimation for digital video

A method for motion coding a digital video data stream comprised of a sequence of pictures having to p and bottom fields, the method including the steps of:

• comparing pixels of a first portion of a current macroblock of the top field of a current picture with pixels of each of a plurality of correspondingly-sized portions of a macroblock of a top field of an anchor picture in accordance with a prescribed search metric, and producing a first error metric for each comparison;
• comparing pixels of the first portion of the current macroblock of the top field of the current picture with pixels of each of the plurality of correspondingly-sized portions of a macroblock of a bottom field of the anchor picture in accordance with the prescribed search metric, and producing a second error metric for each comparison;
• comparing pixels of a second portion of a current macroblock of the bottom field of the current picture with pixels of each of the plurality of correspondingly-sized portions of the macroblock of the top field of the anchor picture in accordance with the prescribed search metric, and producing a third error metric for each comparison;
• comparing pixels of the second portion of the current macroblock of the bottom field of the current picture with pixels of each of the plurality of correspondingly-sized portions of the macroblock of the bottom field of the anchor picture in accordance with the prescribed search metric, and producing a fourth error metric for each comparison;
• summing the first and fourth error metrics to produce a first composite error metric;
• summing the second and third error metrics to produce a second composite error metric;
• and,
• determining which of the first, second, third, and fourth error metrics, and first and second composite error metrics has the lowest value, and selecting one of a plurality of possible motion estimation prediction modes on the basis of such determination.

A method for motion coding a digital video data stream comprised of a sequence of pictures, the method including the steps of:

• comparing pixels of a first portion of a top half of a current macroblock of a current picture with pixels of each of a plurality of correspondingly-sized portions of a macroblock of a top field of an anchor picture in accordance with a prescribed search metric, and producing a first error metric for each comparison;
• comparing pixels of the first portion of the top half of the current macroblock of the current picture with pixels of each of the plurality of correspondingly-sized portions of a macroblock of a bottom field of the anchor picture in accordance with the prescribed search metric, and producing a second error metric for each comparison;
• comparing pixels of a second portion of a bottom half of a current macroblock of the current picture with pixels of each of the plurality of correspondingly-sized portions of the macroblock of the top field of the anchor picture in accordance with the prescribed search metric, and producing a third error metric for each comparison;
• comparing pixels of the second portion of the bottom half of the current macroblock of the current picture with pixels of each of the plurality of correspondingly-sized portions of the macroblock of the bottom field of the anchor picture in accordance with the prescribed search metric, and producing a fourth error metric for each comparison;
• summing the first and third error metrics to produce a first composite error metric;
• summing the second and fourth error metrics to produce a second composite error metric;
• and,
• determining which of the first, second, third, and fourth error metrics, and first and second composite error metrics has the lowest value, and selecting one of a plurality of possible motion estimation prediction modes on the basis of such determination.

A method for motion coding a digital video data stream comprised of a sequence of pictures having top and bottom fields, the method including the steps of:

• comparing pixels of a first portion of a current macroblock of the top field of a current picture with pixels of each of a plurality of correspondingly-sized portions of a macroblock of a top field of an anchor picture in accordance with a prescribed search metric, and producing a first error metric for each comparison;
• comparing pixels of the first portion of the current macroblock of the top field of the current picture with pixels of each of the plurality of correspondingly-sized portions of a macroblock of a bottom field of the anchor picture in accordance with the prescribed search metric, and producing a second error metric for each comparison;
• comparing pixels of a second portion of a current macroblock of the bottom field of the current picture with pixels of each of the plurality of correspondingly-sized portions of the macroblock of the top field of the anchor picture in accordance with the prescribed search metric, and producing a third error metric for each comparison;
• comparing pixels of the second portion of the current macroblock of the bottom field of the current picture with pixels of each of the plurality of correspondingly-sized portions of the macroblock of the bottom field of the anchor picture in accordance with the prescribed search metric, and producing a fourth error metric for each comparison;
• producing first, second, third, and fourth motion vectors on the basis of the first, second, third, and fourth error metrics, respectively;
• and,
• examining the first, second, third, and fourth motion vectors to determine whether a prescribed relationship between them is present, and if so, selecting a frame picture dual-prime motion estimation prediction mode.

A method for motion coding a digital video data stream comprised of a sequence of pictures, the method including the steps of:

• comparing pixels of a first portion of a top half of a current macroblock of a current picture with pixels of each of a plurality of correspondingly-sized portions of a macroblock of a top field of an anchor picture in accordance with a prescribed search metric, and producing a first error metric for each comparison;
• comparing pixels of the first portion of the top half of the current macroblock of the current picture with pixels of each of the plurality of correspondingly-sized portions of a macroblock of a bottom field of the anchor picture in accordance with the prescribed search metric, and producing a second error metric for each comparison;
• comparing pixels of a second portion of a bottom half of a current macroblock of the current picture with pixels of each of the plurality of correspondingly-sized portions of the macroblock of the top field of the anchor picture in accordance with the prescribed search metric, and producing a third error metric for each comparison;
• comparing pixels of the second portion of the bottom half of the current macroblock of the current picture with pixels of each of the plurality of correspondingly-sized portions of the macroblock of the bottom field of the anchor picture in accordance with the prescribed search metric, and producing a fourth error metric for each comparison;
• summing the first and third error metrics to produce a first composite error metric;
• summing the second and fourth error metrics to produce a second composite error metric;
• producing first and second motion vectors on the basis of the first and second composite error metrics, respectively;
• and,
• examining the first and second motion vectors to determine whether a prescribed relationship between them is present, and if so, selecting a field picture dual-prime motion estimation prediction mode.

A method, including the steps of:

• concurrently performing motion estimation using each of a plurality of different motion estimation prediction modes;
• comparing the results produced using each different prediction mode;
• and,
• selecting the prediction mode that produced an optimum result;
• and,
• generating one or more motion vectors using the selected prediction mode.