# US 8,542,935 B2 · Hybrid prediction apparatus and method for entropy encoding

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 8,542,935 B2
- Document ID: US8542935B2
- Status: Granted
- Kind: B2
- Application number: 12/953,163
- Filed: 2010-11-23
- Published: 2011-06-02
- Granted: 2013-09-24
- Earliest claimed: 2009-11-30
- Assignees: Electronics and Telecommunications Research Institute
- Applicants: Kyeongpyo Kim; Woo Yong Lee; Young Ro Kim; HyunSang Park
- Inventors: Kyeongpyo Kim; Woo Yong Lee; Young Ro Kim; HyunSang Park
- Web page: https://patentgrep.com/patent/US8542935B2

## CPC

- H04N 19/91: Entropy coding, e.g. variable length coding [VLC] or arithmetic coding
- H04N 19/593: involving spatial prediction techniques
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION

## Abstract

Provided are a hybrid prediction apparatus and method for entropy encoding that may enhance an existing image compression scheme and prediction scheme by including and selectively using a plurality of predictors configured to perform a per-pixel prediction of an image frame, and may also supplement a performance of a prediction scheme, excessively occurring in a particular pixel.

## Claims

### Claim 1

A hybrid prediction apparatus for entropy encoding, the apparatus comprising:
at least two predictors, each configured to output a plurality of prediction patterns with respect to an input image frame using correlations between pixels of the image frame;
a selection processor to acquire per-pixel errors of all pixels of the image frame for each of the predictors with respect to all prediction pattern combinations, each prediction pattern combination including one of the plurality of prediction patterns output by each of the predictors, and to thereby select, from among all of the prediction pattern combinations for all of the predictors, one prediction pattern combination and one predictor that have a relatively small amount of per-pixel errors; and
an entropy coder to perform entropy encoding of the selected prediction pattern combination.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein the at least two predictors includes two predictors that respectively output the plurality of prediction patterns of the image frame using a median edge detector (MED) scheme and a gradient-adjusted predictor (GAP) scheme.

### Claim 3 (depends on 1)

The apparatus of claim 1, wherein a per-pixel error for each of the pixels of the image frame corresponds to a sum of an absolute difference (SAD) between a prediction value of said each pixel and an actual pixel value of said each pixel.

### Claim 4 (depends on 1)

The apparatus of claim 1, wherein the selection processor comprises:
a counting unit to count a predetermined value through a comparison for the at least two predictors by acquiring the per-pixel errors of the prediction patterns; and
a selector to select the prediction pattern having the relatively small amount of errors from among prediction patterns with respect to the same pixel, using a counting result of the counting unit.

### Claim 5 (depends on 4)

The apparatus of claim 4, wherein the counting unit compares the per-pixel errors of the prediction patterns output by the at least two predictors to increase a count value of a prediction pattern of a predictor having a relatively small per-pixel error.

### Claim 6

A hybrid prediction method of a hybrid prediction apparatus comprising at least two predictors, the method comprising:
predicting, by each of the at least two predictors, a plurality of prediction patterns with respect to an input image frame using correlations between pixels of the image frame;
analyzing per-pixel errors of all pixels of the image frame for each of the predictors with respect to all prediction pattern combinations, each prediction pattern combination including one of the plurality of prediction patterns output by each of the at least two predictors, and selecting, from among all of the prediction pattern combinations and all of the predictors, a prediction pattern combination and a predictor that have a relatively small amount of errors; and
performing entropy encoding of the image frame using the selected prediction pattern combination.

### Claim 7 (depends on 6)

The method of claim 6, wherein the predicting comprises outputting the prediction pattern of the image frame using a MED scheme and a GAP scheme.

### Claim 8 (depends on 6)

The method of claim 6, wherein a per-pixel error for each of the pixels of the image frame corresponds to a sum of an SAD between a prediction value of said each pixel and an actual pixel value of said each pixel.

### Claim 9 (depends on 6)

The method of claim 6, wherein the analyzing and the selecting comprises:
counting a predetermined value through a comparison for the at least two predictors by acquiring the per-pixel errors of the prediction patterns; and
selecting the prediction pattern having the relatively small amount of errors from among prediction patterns with respect to the same pixel, using the count value.

### Claim 10 (depends on 8)

The method of claim 8, wherein the counting comprises comparing the per-pixel errors of the prediction patterns output by the at least two predictors to increase a count value of a prediction pattern of a predictor having a relatively small per-pixel error.