Method and device for binary segmentation of a signal representing a digital image with a threshold obtained using a histogram

• US 5,832,111 A
• Filed: 07/30/1996
• Issued: 11/03/1998
• Est. Priority Date: 12/21/1990
• Status: Expired due to Fees

• Current Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Original Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Inventors: Florent, Raoul
• Primary Examiner(s): Mancuso, Joseph
• Assistant Examiner(s): Prikockis, Larry J.
• Application Number:
  US08/688,580
• Time in Patent Office: 826 Days
• Field of Search: 382/270, 382/271, 382/273, 382/192, 382/173, 382/174, 382/204, 382/168, 382/171, 382/172, 358/465, 358/466
• US Class Current: 382/171
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/136 involving thresholding
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns

Abstract

Method and device for binary segmentation of a digital image with a threshold obtained using a histogram to obtain a binary segmented digital image. The digital image is thresholded a number of grey levels, and the number of objects which appear when the digital image is thresholded at each of the grey levels is counted. A histogram of the number of objects per grey level is constructed, and a segmentation threshold is determined automatically on the basis of the histogram. The digital image is then thresholded at the segmentation threshold to obtain the binary segmented digital image.

Claims

1. A method of binary segmentation of a video image which includes at least one target object and at least one noise element against a common background;

said image being represented by a video signal corresponding to grey levels of respective pixels therein, a target object having pixel grey levels which differ from the pixel grey levels of the background and of noise elements;

said method comprising the steps of;

(a) thresholding the video signal at respective grey threshold levels so as to derive for each threshold level a digital signal representing a binary segmented image corresponding to that threshold level;

(b) determining from each of the digital signals, by labelling, a count of the number of objects in the corresponding binary segmented image, and storing the respective counts applicable to the respective binary segmented images;

(c) forming a histogram of the stored counts as a function of the respective threshold levels;

(d) automatically determining from said histogram a segmentation threshold applicable to said video image for distinguishing target objects from background and noise elements therein; and

(e) applying said segmentation threshold to said video signal to derive a binary segmented video signal representing a binary segmented form of said video image in which target objects are distinguished from background and noise elements therein.

2. The method according to claim 1, wherein said segmentation threshold is determined by locating a first grey level above which the count of said histogram never exceeds a predetermined number.

3. The method according to claim 1, wherein said segmentation threshold is determined by locating the lowest grey level in a range of grey levels for which said histogram has a flat, non-zero height.

4. The method according to claim 1, further comprising smoothing said histogram to form a smoothed histogram curve;

and wherein said segmentation threshold is determined by

determining, for each of a plurality of test grey levels, a deviation measure combining a deviation of a first slope of said histogram curve from a vertical line and a deviation of a second slope of said histogram curve from a horizontal line; and

locating that one of said test grey levels having the minimum for said deviation measure.

5. The method according to claim 4, wherein said deviation measure is a sum of the squares of the deviations of said first and second slopes.

6. The method according to claim 2, wherein said predetermined number is the number of target objects expected in video said image.

7. The method according to claim 1, further comprising smoothing said histogram to form a smoothed histogram curve;

and wherein said segmentation threshold is determined by locating a first grey level above which the count of said histogram curve never exceeds a predetermined number.

8. The method according to claim 7, wherein said predetermined number is the number of target objects expected in said video image.

9. The method according to claim 1, further comprising smoothing said histogram to form a smoothed histogram curve;

and wherein said segmentation threshold is determined by locating a lowest grey level in a highest range of grey levels in which said histogram curve has a flat, non-zero height.

10. The method according to claim 1, further comprising processing said video signal prior to thresholding thereof so as to compensate for grey level variations of the background of said video image but without degrading representation by said video signal of contours of said at least one target object.

11. Apparatus for binary segmentation of a video image which includes at least one target object and at least one noise element against a common background;

said image being represented by a video signal corresponding to grey levels of respective pixels therein, a target object having pixel grey levels which differ from the pixel grey levels of the background and of noise elements;

said apparatus comprising;

input means for receiving the video signal;

thresholding means coupled to said input means for subjecting the video signal to thresholding at respective grey threshold levels so as to derive for each threshold level a digital signal representing a binary segmented image corresponding to such threshold level;

processing means coupled to said thresholding means for determining from each digital signal, by labelling, a count of the number of objects in the corresponding binary segmented image;

accumulating means coupled to said processing means for storing the respective counts applicable to the respective binary segmented images;

segmentation threshold determining means coupled to said accumulating means for automatically determining a segmentation threshold on the basis of said stored counts; and

means coupled to said segmentation threshold determining means and to said input means for thresholding said video signal in accordance with said segmentation threshold, to thereby derive a binary segmented video signal representing a binary segmented form of said video image wherein target objects are distinguished from background and noise elements therein.

12. Apparatus according to claim 11, wherein said video image is a localized portion of a larger video image and said apparatus further comprises means for deriving a video signal representative of said localized portion from a video signal which is representative of said larger video image.

13. Apparatus according to claim 11, further comprising top hat filter means coupled to said input means for equalizing said video signal for grey level variations of the background of said video image but without degrading representation by said video signal of contours of said at least one target object.

14. Video image segmentation apparatus 1, wherein said segmentation thresholding determining means comprises:

means for creating a representation of a smooth histogram curve of said objects as a function of said respective grey levels; and

means for determining a grey level at which a first slope of said histogram curve approaches infinity and a second slope of said histogram curve approaches zero.