Automatic threshold generation technique

[54] AUTOMATIC THRESHOLD GENERATION TECHNIQUE

[75] Inventors: Mircea P. Marandici, Poughkeepsie, N.Y.; Matthew J. Quinn, Seattle, Wash.

[73] Assignee: International Business Machines Corporation, Armonk, N.Y.

[21] Appl. No.: 100,637

[22] Filed: Jul. 30, 1993

Related U.S. Application Data

[63] Continuation of Ser. No. 783,453, Oct. 24, 1991, abandoned.

[51] Int. C1.5 G06K 9/38

[52] U.S. CI 382/51; 358/466

[58] Field of Search 382/9, 18, 50, 51, 52,

382/53; 358/464, 465, 466

[56] References Cited

U.S. PATENT DOCUMENTS

3,944,977 3/1976 Holmes 340/146.3

4,064,484 12/1977 Mese et al 340/146.3

4,442,544 4/1984 Moreland et al 382/53

4.468.703 8/1984 Fujiwara et al 340/146.3

4.468.704 8/1984 Stoffel et al 382/52

4,593,325 6/1986 Kannapell et al 382/50

4,656,665 4/1987 Pennebaker 382/18

4,710,822 12/1987 Matsunawa 382/51

4.742.556 5/1988 Davis, Jr. et al 382/51

4.742.557 5/1988 Ma 382/51

4,907,286 3/1990 Yamada 382/51

4,977,605 12/1990 Fardeau et al 382/51

5,075,872 12/1991 Kumagai 382/51

5,138,671 8/1992 Yokoyama 358/466

5,179,599 1/1993 Formanek 382/51

OTHER PUBLICATIONS

IBM Technical Disclosure Bulletins, vol. 15, No. 8, pp. 2595-2598 (two articles), Jan. 1973. IBM Technical Disclosure Bulletin, vol. 18, No. 6, pp. 1962-1965, Nov. 1975.

IBM Technical Disclosure Bulletin, vol. 5, No. 6, pp. 55-56, Nov. 1962.

IBM Technical Disclosure Bulletin, vol. 22, No. 3, pp. 1031-1033, Aug. 1979.

IBM Technical Disclosure Bulletin, vol. 21, No. 7, pp. 3001-3003, Dec. 1978.

IBM Technical Disclosure Bulletin, vol. 15, No. 4, pp. 1138-1139, Sep. 1972.

Computer Science Press, "Algorithms for Graphics and Image Processing", Theo Pavlidis (cover sheet and page 66, only).

Primary Examiner—Michael T. Razavi

Assistant Examiner—David Fox

Attorney, Agent, or Firm—Whitham, Curtis & Whitham

An estimate of a threshold value for feature extraction from image data containing noise and quantized into a plurality of quantization levels is made by filtering, such as recursively averaging n adjacent quantization levels and counting the extrema of the result. The recursive averaging is repeated with increasing values of n until only three extrema are present in the filtered data. The estimate of the threshold value is refined by comparing averages or summations of n quantization levels above and below the estimate and altering the estimated threshold value based on a comparison of the averages or summations. By basing both the estimation and refinement processes on the original data and using a value of n which is smaller in the refinement process than in the estimation process, a highly effective threshold value for the feature extraction process is obtained. Adaptive adjustment of n during the estimation process minimizes loss of data as well as computing time.

12 Claims, 5 Drawing Sheets

205 r

BUFFER

2B-7"

233