Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Predictor Type Determination Algorithm for JPEG Lossless Process

IP.com Disclosure Number: IPCOM000115623D
Original Publication Date: 1995-Jun-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 2 page(s) / 33K

Publishing Venue

IBM

Related People

Ishii, H: AUTHOR

Abstract

Disclosed is the heuristic algorithm to determine the predictor type of JPEG (*) lossless process. Currently there is no method to select the optimum predictor type except the exhaustive search for all image data. But because of the heuristic method, this predictor determination algorithm is faster than the exhaustive search several hundreds times to determine the predictor type.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 96% of the total text.

Predictor Type Determination Algorithm for JPEG Lossless Process

      Disclosed is the heuristic algorithm to determine the predictor
type of JPEG (*) lossless process.  Currently there is no method to
select the optimum predictor type except the exhaustive search for
all image data.  But because of the heuristic method, this predictor
determination algorithm is faster than the exhaustive search several
hundreds times to determine the predictor type.

      This algorithm consists of 2 phase, and the over view of this
algorithm follows:
  < Phase 1 >
  Sample the pixel value of input raw image data.  (Sampling rate is
   5%.)
  < Phase 2 >
  Calculate the estimate function for sampling pixel value and
   select the predictor type whose estimate function result is
smallest.
  Estimate function is
    Fj = E x Ra / Rj
  where
  j = j is the predictor type 1,2,...,7.
  E = error sum of modulus
    (error is the difference between pixel value and predictor
result)
  Ra = average of the number of times in which the predictor result
   is equal to the pixel value for all predictor type
  Rj = average of the number of times in which the predictor result
   is equal to the pixel value for each predictor type.

      For image process system, a calculation time is most important
factor, so this algorithm is effective from the view point of process
time and compression efficiency.
  Reference
  (*) "Final Text of DIS 10918-1: Digital Compress...