Trap filter

An adaptive filter system comprising:

• a learning rate component that stores a learning rate parameter;
• an observed value component that provides an observed values of coefficients;
• a previously coded values component that stores previously computed probabilities;
• and
• an estimator that generates probabilities for the coefficients as a function of a learning rate parameter, the observed values, the previously computed probabilities and distances of the previously computed probabilities.

An adaptive filter system comprising:

• a first filter that processes lines of an image, respective lines comprising coefficients, the coefficients having a probability computed as a function of a probability of a previous pixel;
• a second filter that processes columns of the image, the probability of coefficients being updated as a function of the probability of above pixel;
• and
• a third filter that processes the lines of the image after the coefficients of that line have been processed, and in an opposite direction of the first filter, the probability of coefficients being updated as a function of the previous pixel.

An encoding system comprising:

• a tarp filter that generates a probability of a current coefficient;
• a context predictor that generates a probability of a current coefficient;
• a select component that selects one of the probability generated by the tarp filter and the probability generated by the context predictor as a selected probability;
• and
• an encoder that encodes the current coefficient utilizing the selected probability.

A method of encoding a document comprising:

• transforming pixels of an image into coefficients;
• quantizing the coefficients;
• generating probabilities for the coefficients based at least partly a subset of previously computed probabilities of previously encoded coefficients and distances of the previously encoded coefficients;
• and
• encoding the coefficients into a bitstream by utilizing the generated probabilities.

A method of operating a tarp filter comprising:

• determining a learning rate parameter;
• receiving at least one previously computed probability;
• and
• computing a probability of the current coefficient based on the at least one previously computed probability and a distance of the at least one previously computed probability.

A computer readable medium storing computer executable instructions of components operable to perform a method of encoding, comprising:

• a component for transforming pixels of an image into coefficients;
• a component for quantizing the coefficients;
• a component for generating probabilities for the coefficients based at least partly on substantially all previously computed probabilities of previously encoded coefficients and distances of substantially all previously computed probabilities;
• and
• a component for encoding the coefficients into a bitstream by utilizing the generated probabilities.

A computer readable medium storing computer executable instructions of components operable to generate probabilities comprising:

• a component for determining a learning rate parameter;
• a component for receiving at least one previously computed probability;
• and
• a component for computing a probability of the current coefficient based on the at least one previously computed probability and a distance of the at least one previously computed probability.

A system comprising:

• an decoding system comprising:
• a probability estimator that adaptively computes probabilities for transform coefficients of an image, the probabilities being a function of probabilities of previously decoded transform coefficients and a decreasing function of a distance of previously decoded transform coefficients;
• and
• an entropy decoder that decodes transform coefficients utilizing the probabilities.