Appeal No. 1997-4067 Application 08/285,324 line 3). Thus, Williams discloses step (a) and discloses steps (b) and (c) except for the last limitation of "generating an error and binary value therefrom." Williams also fails to disclose steps (d) and (e). The Examiner relies on Kanno. Kanno discloses binarizing a multi-level pixel (B output at 41), generating an error therefrom (EB), and using error diffusion. Kanno notes that a maximum density difference of pixels within a predetermined 4x4 area surrounding the pixel of interest is large for a character (text) region having a high contrast and is small for a photograph region having a gray scale (col. 2, lines 38-42). Kanno uses the average maximum density difference signal (AD) to adjust quantization of the pixel of interest to either the quantization error (EB) (in a photograph region) or zero (in a text region) (col. 5, lines 55-60). When the compensation error signal (CE) is not zero (i.e., for a photograph region), weighting coefficients from the weighting coefficient memory 7 are used to distribute the error to surrounding pixels (col. 6, lines 12-30). The weighting coefficient memory 7 contains only a single set of weighting coefficients. Thus, when a - 11 -Page: Previous 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 NextLast modified: November 3, 2007