Image processor with smooth transitioning between dither and diffusion processes

• US 5,014,333 A
• Filed: 07/21/1988
• Issued: 05/07/1991
• Est. Priority Date: 07/21/1988
• Status: Expired due to Fees

• Current Assignee: Eastman Kodak Company
• Original Assignee: Eastman Kodak Company
• Inventors: Miller, Rodney L., Morton, Roger R. A.
• Primary Examiner(s): Moore, David K.
• Assistant Examiner(s): Mancuso, Joseph
• Application Number:
  US07/222,610
• Time in Patent Office: 1,020 Days
• Field of Search: 382/50, 382/52, 382/54, 358/443, 358/453, 358/463, 358/464, 358/261.2, 358/466
• US Class Current: 382/252
• CPC Class Codes: H04N 1/4053 with threshold modulated relative to input image data or vice versa

Abstract

An image processor converts a multiple gray-level image to a bi-tonal image using both error diffusion and ordered dither to enhance image quality. The error diffusion process is attenuated smoothly as the local area intensity approaches a light or dark maximum, while the ordered dither process is attenuated smoothly as the local area high spatial frequency content increases.

Claims

1. An image processor comprising:

image memory means for storing an image comprising a frame of M columns and N rows of multi-gray level pixels;

adder means for forming successive input amplitudes from successive ones of said pixel so as to form a current input amplitude corresponding to the current one of said pixels;

means for comparing said current input amplitude with a dithered threshold so as to generate;

a. a binary print/no print decision,

b. a feedback error;

means for attenuating said feedback error so as to produce an attenuated error, said attenuating means responsive to the average gray level of a local neighborhood of pixels surrounding said current pixel so as to attenuate said feedback error proportionately as said average gray level approaches a light or dark limit;

means for transmitting said attenuated error to said adder means so that said adder means adds said attenuated error to the next one of said successive pixels to form the next one of said successive input amplitudes; and

means for transmitting said dithered threshold to said comparator means, comprising;

a. means for storing a plurality of ordered dither matrices, each of said matrices defining plural patterns of dithered thresholds as individual elements in each of said matrices, each of said patterns characterized by dither threshold amplitudes for each of said M ×

N pixels spatially modulated with a contour of a characteristic sharpness,

b. means for sensing the amplitude of the high spatial frequency content of said local neighborhood,

c. means for selecting one of said plurality of ordered dither matrices whereby the characteristic sharpness of the one ordered dither matrix so selected is inversely proportional to the amplitude of the high spatial frequency content of said local neighborhood, and

d. means for selecting for said dithered threshold an element within the one selected ordered dither matrix corresponding to the position of the current one of said successive pixels stored in said image storing means, whereby an ordered dither process is provided which is attenuated in proportion to the amplitude of the high spatial frequency content of the local neighborhood surrounding each successive one of said M ×

N pixels.

2. The processor of claim 1, wherein said means for attenuating said feedback error comprise:

low pass filter means for determining said average gray level of said local neighborhood of pixels surrounding said current pixel, said local neighborhood comprising a kernel of pixels on the order of about five rows and five columns of such pixels surrounding said current pixel;

converter means responsive to said high pass filter means for generating an attenuation factor which decreases as said average gray level of said local neighborhood approaches either one of said light or dark limit;

means for multiplying said feedback error by said attenuation factor.

3. The processor of claim 2 wherein said attenuation factor is one of a set of discreet attenuation factors lying in a range between a minimum attenuation factor and a maximum attenuation factor, said processor further comprising means for dividing by said maximum attenuation factor the product produced by said means for multiplying so as to produce said attenuated error.

4. The processor of claim 2 wherein said converter means comprises a read only memory storing a set of discreet attenuation factors each being addressable within said read only memory by a corresponding value received from said low pass filter.

5. The processor of claim 1 wherein said means for sensing the amplitude of said high spatial frequency content of said local neighborhood comprises a high pass filter characterized by a kernel of on the order of about five rows and five columns of said pixels comprising said local neighborhood surrounding said current pixel, said high pass filter adaptable to generate a high pass amplitude word comprising L bits.

6. The processor of claim 5 wherein said means for selecting said element within the one selected ordered dither matrix comprise a read only memory, wherein said read only memory is addressable only by the K most significant bits of said L bit word generated by said high pass filter, and wherein said read only memory stores 2^K ordered dither matrices.

7. The processor of claim 6 wherein said read only memory is further addressable by one of N×

M location words corresponding to the location of said current pixel in said image memory means, whereby said one location word selects within said read only memory the corresponding element in the selected one of said 2^K ordered dither matrices

8. The processor of claim 1 wherein said feedback error comprises (a) the difference between said input amplitude and a known print value if said print/no print decision comprises a decision to print, or (b) the difference between said input amplitude and a value corresponding to zero if said print/no print decision comprises a decision to not print.

9. An image processor for receiving a multiple gray level image comprising an array of image pixels and producing therefrom a bitonal image, said image processor comprising:

error diffusion means for diffusing among neighboring pixels the difference between each image pixel in said array and a binarization threshold so as to produce a succession of error diffused pixels;

ordered dither means for dithering said binarization threshold in accordance with a spatially variant ordered dither pattern of threshold values dithered in accordance with a given contour having a characteristic amplitude excursion;

first attenuation means for attenuating the difference diffused among said neighboring pixels by said error diffusion means proportionally as the average value of a neighborhood of pixels surrounding a corresponding image pixel approaches a light of dark limit; and

second attenuation means for attenuating the amplitude excursion of said ordered dither pattern in proportion to the amplitude of the high spatial frequency content of said neighborhood of pixels.