Processing of gray scale image pels to facilitate data compression

• US 4,599,656 A
• Filed: 02/21/1984
• Issued: 07/08/1986
• Est. Priority Date: 02/21/1984
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Bellinghausen, Joan M.
• Primary Examiner(s): Orsino, Jr., Joseph A.
• Application Number:
  US06/581,768
• Time in Patent Office: 868 Days
• Field of Search: 382/50, 382/52, 382/54, 382/56, 358/166, 358/284, 358/260, 358/261, 358/280, 358/21 R, 358/36, 358/167
• US Class Current: 382/237
• CPC Class Codes: G06T 3/403 Edge-driven scaling; Edge-based scaling
• CPC Class Codes: H04N 1/403 Discrimination between the two tones in the picture signal of a two-tone original
• CPC Class Codes: H04N 1/409 Edge or detail enhancement; Noise or error suppression

Abstract

A process is described for identifying those pels in a gray scale image which are candidates for vertical filtering before the image is thresholded to binary or black/white pel values. The vertical neighbor pels of a pel below a run end point are examined to determine whether their gray scale values fall within the same range. If they do, the horizontal neighbors of the current pel are examined to see whether their gray scale values fall in different ranges. If they do, the pel is vertically filtered by adjusting its gray scale value as a function of the differences between its value and the values of its vertical neighbors. Because only selected pels are vertically filtered, the process can be carried out with a general purpose microprocessor rather than a special purpose parallel processor.

Claims

1. For use in an image processing system of the type wherein capture gray scale pel values are to be thresholded to binary values to facilitate subsequent data compression, a method for identifying a pel which may require filtering to reduce distortion due to sampling along image vertical edges, said method comprising the steps of:

retrieving capture gray scale pel values for a current pel and for both pels which are vertical neighbors of the current pel;

assigning each vertical neighbor pel one of a limited number of gray scale range values as a function of the capture gray scale value of the pel;

identifying, as a filtering candidate, any current pel having vertical neighbors with the same assigned value;

retrieving capture gray scale pel values for both pels which are horizontal neighbors of each candidate current pel;

assigning each horizontal neighbor pel one of a limited number of gray scale range values as a function of the capture gray scale value of the pel; and

finally selecting, for filtering, each candidate current pel having horizontal neighbors with different assigned values.

2. A method for vertically filtering any current pel identified by the steps of claim 1 comprising the additional step of adjusting the gray scale value of any identified current pel as a function of the relative gray scale values of the identified current pel and each of its vertical neighbors.

3. A method for vertically filtering identified current pels as set forth in claim 2 comprising the steps of:

establishing the difference in capture gray scale values of the current pel and its upper vertical neighbors;

establishing the difference in capture gray scale value of the current pel and its lower vertical neighbor;

adjusting the gray scale value of the current pel as a function of the established differences.

4. A method for compressing gray scale pel data, including data vertically filtered in accordance with claim 3, comprising the steps of:

assigning one of two binary values to each pel as a function of the filtered gray scale values of the identified current pels and the unfiltered gray scale values of all remaining pels; and

run length compressing successive lines of pels.

5. For use in an image processing system of the type wherein capture gray scale pel values are to be thresholded to binary values to facilitate subsequent data compression, a method for processing selected pels which may require filtering to reduce distortion due to sampling along image vertical edges, said method comprising the steps of:

(a) capturing and storing gray scale pel values for an entire image field in a multi-line frame buffer;

(b) thresholding the stored gray scale pel values in the first image line to establish a run length history for that line;

(c) using the run length history of the prior line to locate the pel below a run endpoint;

(d) retrieving capture gray scale pel values for the located pel and for both pels which are vertical neighbors of the located pel;

(e) assigning each vertical neighbor pel one of a limited number of gray scale range values as a function of its capture gray scale value;

(f) identifying, as a filtering candidate, any located pel having vertical neighbors with the same assigned value;

(g) retrieving capture gray scale pel values for pels which are horizontal neighbors to each identified pel;

(h) assigning each horizontal neighbor pel one of a limited number of gray scale range values as a function of its capture gray scale value;

(i) vertically filtering any candidate pel having horizontal neighbor pels with different assigned values;

(j) repeating steps (d) through (i) for the pel to the right of the originally located pel;

(k) repeating steps (c) through (j) for the remaining run endpoints indicated by the run length history;

(l) thresholding the gray scale pel values for the just-processed line to establish a run length history for that line; and

(m) repeating steps (c) through (k) using the just-established run length history for all image lines in the frame buffer except the last line.