Guaranteed-rate tiled image data compression

1. A method of compressing digital image data, the method comprising:

a. for each image data block in a plurality of image data blocks in the digital image data,

splitting the image data in the image data block into luminance data and chrominance data;

transforming the luminance data using a two-dimensional integer wavelet transform to convert the luminance data to a low-frequency luminance coefficient and a plurality of high-frequency luminance coefficients;

transforming the chrominance data using a two-dimensional integer wavelet transform to convert the chrominance data to a low-frequency chrominance coefficient and a plurality of high-frequency chrominance coefficients;

computing a predicted low-frequency luminance coefficient for the luminance data based on at least one neighboring image data block in the plurality of image data blocks;

computing a predicted low-frequency luminance coefficient for the chrominance data based on at least one neighboring image data block in the plurality of image data blocks;

computing a residual luminance low-frequency coefficient based on the predicted luminance low-frequency coefficient and the low-frequency luminance coefficient;

computing a residual chrominance low-frequency coefficient based on the predicted chrominance low-frequency coefficient and the low-frequency chrominance coefficient;

selecting a quantization parameter, a luminance entropy coding parameter, and a chrominance entropy coding parameter using rate control, wherein the rate control guarantees a predetermined compression rate;

quantizing the plurality of high-frequency luminance coefficients and the plurality of high-frequency chrominance coefficients based on the quantization parameter;

entropy coding the residual low-frequency luminance coefficient and the quantized high-frequency luminance coefficients based on the luminance entropy coding parameter; and

entropy coding the residual low-frequency chrominance coefficient and the quantized high-frequency chrominance coefficients based on the chrominance entropy coding parameter.

2. The method of claim 1 , wherein the rate control allocates an extra initial bit-rate to an initial image data block and gradually reduces the extra initial bit-rate for subsequent image data blocks to guarantee the predetermined compression rate.

3. The method of claim 1 , wherein

transforming the luminance data comprises performing a sufficient number of lifting steps are performed to convert the luminance data into a single low-frequency luminance coefficient and the plurality of high-frequency luminance coefficients; and

transforming the chrominance data comprises performing a sufficient number of lifting steps to convert the chrominance data into a single low-frequency chrominance coefficient and the plurality of high-frequency chrominance coefficients.

4. The method of claim 1 , wherein

computing a predicted low-frequency luminance coefficient comprises computing the predicted low-frequency luminance coefficient based on a left neighboring image data block, a top neighboring image data block, and a top-left neighboring image data block; and

computing a predicted low-frequency chrominance coefficient comprises computing the predicted low-frequency chrominance coefficient based on a left neighboring image data block, a top neighboring image data block, and a top-left neighboring image data block.

5. The method of claim 1 , further comprising coding the quantization parameter, the luminance entropy coding parameter, and the chrominance entropy coding parameter.

6. The method of claim 1 , wherein the digital image data is a reference frame.