United States Patent m
 METHOD AND APPARATUS FOR
PROVIDING BIT-RATE REDUCTION AND
RECONSTRUCTION OF IMAGE DATA
USING DITHER ARRAYS
 Inventor: Yigal Accad, Millbrae, Calif.
 Assignee: Electronics For Imaging, Inc., San
 Appl. No.: 398,369
 Filed: Mar. 3,1995
 Int. CI.6 H04N 1/405; H04N 1/413
 U.S. CI 395/109; 395/114; 358/457;
 Field of Search 395/109, 114;
358/457, 429, 426, 456, 466, 465; 382/244,
237, 270, 251
 References Cited
U.S. PATENT DOCUMENTS
3,244,808 4/1966 Roberts .
4,667,251 5/1987 Hasegawa 358/457
5,111,310 5/1992 Parker etal 358/457
5,323,247 6/1994 Parker etal 358/457
5,341,228 8/1994 Parker et al 358/457
5,477,305 12/1995 Parker et al 358/457
1992 Optical Society of America, J. Opt. Soc. Am. A/vol 9 No. 11, Nov. 1992, Mitsa, et al., "Digital Halftoning Technique Using A Blue-Noise Mask".
i urn iimin in Iiiii urn imi Ihi Iiiii Iiiii Iiiii Iiiii Huh Hi Iiiii Iiii
US005553200A [ii] Patent Number: 5,553,200  Date of Patent: Sep. 3,1996
IRE Transactions on Information Theory, Roberts, "Picture Coding Using Pseudo Random Noise," Feb. 1962, IT-8, No. 2, pp. 145-154.
"The Void-And-Cluster Method For Dither Array Generation," by Robert Ulichney, Sep. 1993, Proceedings of the SPIE, vol. 1913, pp. 332-343.
Primary Examiner—Scott A. Rogers
Attorney, Agent, or Firm—Hecker & Harriman
An apparatus and method for providing bit-rate reduction and reconstruction of image data. Bit-rate reduction of 32-bit CMYK combinations into 16-bit code words is performed on a pixel by pixel basis by stepwise calculations or by using LUTs. Bit-rate reduction takes place during PostScript® interpretation where the output image pixels are generated in a possibly arbitrary order. Bit expansion from 16-bit code words to 32-bit data for 8 bits per component of the CMYK image values is performed while real time printing, preferably by direct table look up. During bit-rate reduction, the image data may be non-linearly corrected to compensate for perceptual non-uniformities and for nonlinearities in the input/output relationship of the reproduction device. Quantization and coding may be accomplished by thresholding the non-linearly corrected image data by values from a dither array. The coded image is stored in the frame buffer and subsequently retrieved, de-dithered, dequantized and inversely transformed for the perceptual non-linearity before printing.
24 Claims, 4 Drawing Sheets