Method and apparatus for image compression

An encoding apparatus, comprising:

• a memory;
• and
• a processor that is in communication with the memory and has a structure to perform instructions from a program stored in the memory so that the following image compression process involving wavelet transformation of an image, wherein subband image samples are formed into code-block, is performed:
• determining for each of the subband image samples whether the subband image sample is significant or insignificant, wherein a subband image sample is determined to be significant if the subband image is non-zero and wherein a subband image sample is determined to be insignificant otherwise;
• for each subband image sample determined to be significant, computing a magnitude-minus-one value for the subband image sample;
• for each subband image sample, computing a magnitude exponent for the subband image sample, the magnitude exponent being a smallest non-negative integer e, such that 2 raised to the power of e is greater than twice the magnitude-minus-one;
• collecting the subband image sample of the code-block into groups of n consecutive subband image samples in a scanning order in accordance with a scanning pattern;
• assigning a coding context to each of the groups, the coding context depending only on the significance of subband image samples associated with previous groups in scanning order, wherein groups are identified as all zero content (AZC) groups, wherein all of the subband image samples used to from the coding context for the group are insignificant;
• encoding significance information on a group by group basis with at least two separate bit-streams, the two separate bit-streams consisting of:
• a first bit-stream that encodes a group significance symbol for all groups identified as AZC groups, wherein the group significance symbol indicates whether or not any subband image sample in the group is significant, and
• a second bit-stream that jointly encodes the significance of all subband image samples in a group, for non-AZC groups and significant AZC groups, wherein at least one subband image sample in the group is significant, and wherein the at least two separate bit-streams being able to be concurrently decoded;
• and
• coding the magnitude-minus one value and sign information for significant subband image samples, using codes that depend only on the significance and magnitude-exponent of previous subband image samples in the scanning order;
• arranging the significance and magnitude code bits for the code blocks into the at least two separate bit streams that are able to be decoded concurrently.