Video coding system and method using 3-D discrete wavelet transform and entropy coding with motion information

A computer-implemented method comprising computer-executable instructions, the method comprising:

• estimating motion trajectories of pixels in a video object from frame to frame in a video sequence;
• performing a temporal wavelet transform on the corresponding pixels along the motion trajectories in a temporal direction to produce frames of temporal wavelet coefficients;
• forming a pixel array containing pixels that continue from frame to frame in the video sequence;
• examining a pixel in a frame to determine whether the pixel is a terminating pixel that does not continue to a next frame;
• terminating the pixel array when the pixel is a terminating pixel;
• performing a spatial wavelet transform on the frames of the temporal wavelet coefficients to produce multiple sub-bands of wavelet coefficients;
• coding each sub-band of wavelet coefficients independently;
• coding the wavelet coefficients in bit-planes;
• allocating bits for the bit-planes according to a rate-distortion optimization;
• and
• truncating bits allocated to a bit-plane at a point on a rate-distortion curve that approximates a convex hull.

A computer-implemented method encoding computer-executable instructions, the method comprising:

• transforming frames in a video sequence using a wavelet transform, wherein the transforming comprises:
• forming a pixel array containing pixels that continue from frame to frame in the video sequence;
• examining a pixel in a frame to determine whether the pixel is a terminating pixel that does not continue to a next frame;
• terminating the pixel array when the pixel is a terminating pixel;
• performing a spatial wavelet transform on the frames of the temporal wavelet coefficients to produce multiple sub-bands of wavelet coefficients;
• coding sub-bands of coefficients produced from the transforming in an independent manner such that one sub-band of coefficients is coded independently of another sub-band of coefficients;
• and
• constructing a bitstream from the independently coded sub-bands, wherein the constructing comprises forming multiple bit-planes and truncating a number of bits in each bit-plane according to a rate-distortion curve.

A video encoder comprising:

• a processor coupled to a memory;
• a wavelet transformer stored on the memory and configured to execute on the processor to transform frames in a video sequence into multiple sub-bands of coefficients, the wavelet transform using motion information of video objects in the frames, the transforming comprising:
• forming a pixel array containing pixels that continue from frame to frame in the video sequence;
• examining a pixel in a frame to determine whether the pixel is a terminating pixel that does not continue to a next frame;
• terminating the pixel array when the pixel is a terminating pixel;
• performing a spatial wavelet transform on the frames of the temporal wavelet coefficients to produce multiple sub-bands of wavelet coefficients;
• and
• a coder stored on the memory and configured to execute on the processor to code the coefficients of each sub-band independently, wherein the coder comprises a context-based arithmetic coder to assign contexts to the coefficients of each sub-band based on different coding primitives and the coder truncates a number of coding bits according to rate-distortion curves.

A video encoder comprising:

• means for estimating motion trajectories of pixels in a video object from frame to frame in a video sequence;
• means for performing a temporal wavelet transform on the corresponding pixels along the motion trajectories in a temporal direction to produce frames of temporal wavelet coefficients, the means for performing a temporal wavelet transform comprising:
• means for forming a pixel array containing pixels that continue from frame to frame in the video sequence;
• means for examining a pixel in a frame to determine whether the pixel is a terminating pixel that does not continue to a next frame;
• means for terminating the pixel array when the pixel is a terminating pixel;
• and
• means for adding the pixel to the pixel array when the pixel is not a terminating pixel;
• means for performing a spatial wavelet transform on the frames of the temporal wavelet coefficients to produce multiple sub-bands of wavelet coefficients;
• means for coding each sub-band of wavelet coefficients independently, the coding means comprising:
• means for coding the wavelet coefficients in bit-planes;
• means for allocating bits for the bit-planes according to a rate-distortion optimization;
• and
• means for truncating bits allocated to a bit-plane at a point on a rate-distortion curve that approximates a convex hull.

A computer-readable storage medium encoding computer-executable instructions that, when executed on a processor, direct a device to:

• transform frames in a video sequence using a wavelet transform, wherein the transforming comprises:
• form a pixel array containing pixels that continue from frame to frame in the video sequence;
• examine a pixel in a frame to determine whether the pixel is a terminating pixel that does not continue to a next frame;
• terminate the pixel array when the pixel is a terminating pixel;
• perform a spatial wavelet transform on the frames of the temporal wavelet coefficients to produce multiple sub-bands of wavelet coefficients;
• code sub-bands of coefficients produced from transforming video frames in an independent manner such that one sub-band of coefficients is coded independently of another sub-band of coefficients;
• construct a bitstream from the independently coded sub-bands;
• code the coefficients of each sub-band bit-plane by bit-plane using different coding primitives;
• and
• truncate a number of bits in each bit-plane according to rate-distortion curves.