Program products for iterative quantization rate control in video encoding

摘要

Iterative video encoding systems, methods and computer program products, where residue quantization and data packing operations of an encoding process may he repeated with various values for a quantization parameter, without repeating the determination of macroblock prediction codes. In an embodiment, the size of an actual file generated by encoding is compared to a target file size. The QP may be adjusted depending on the amount by which these file sizes differ. The quantization and packing may then be repeated with the adjusted QP. In an embodiment, a greater difference in these file sizes results in a greater adjustment to the QP.

主权项

1. A video encoding method, comprising:
determining, in a video encoder, one or more macroblock prediction codes for a video frame to be encoded; quantizing residue data remaining from said determination based on a quantization parameter (QP); packing the quantized residue into a bitstream; and evaluating a difference between an actual file size that results from said packing and a target file size; selecting, in response to the actual file size being smaller than the target file size, a QP value decrement from a set of predetermined decrements based on a comparison of the difference to a set of first difference thresholds that are each associated with one of the predetermined decrements, and selecting, in response to the actual file size being larger than the target file size, a QP value increment from a set of predetermined increments based on a comparison of the difference to a set of second difference thresholds that are each associated with one of the predetermined increments, wherein:
the set of first difference thresholds includes at least four thresholds equal to different fractions of the target file size between one-sixteenth and one-half of the target file size;the set of second difference thresholds includes at least four thresholds equal to different fractions of the target file size between one-eighth of the target file size and the target file size; anda magnitude of the QP value increment and decrement increases with a magnitude of the first and second difference thresholds; modifying the value of the QP by the selected QP value decrement or increment; and repeating said quantization, with the modified QP value, and said packing, without repeating said determination.