Method for segmentation of digital color images

• US 5,933,524 A
• Filed: 02/07/1996
• Issued: 08/03/1999
• Est. Priority Date: 09/27/1993
• Status: Expired due to Fees

• Current Assignee: Siemens AG
• Original Assignee: Siemens AG
• Inventors: Schuster, Rolf, Ahmad, Subutai
• Primary Examiner(s): Coles, Edward L.
• Assistant Examiner(s): Lee, Cheukfan
• Application Number:
  US08/591,510
• Time in Patent Office: 1,273 Days
• Field of Search: 382/170, 382/171, 382/199, 382/190, 382/168, 382/166, 382/165, 382/167, 382/162, 382/173, 382/164, 382/224, 382/225, 358/522, 358/538, 358/518
• US Class Current: 382/168
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation

Abstract

Colored objects are represented by digitized color histograms which are used for the segmentation of digital color images. In this case, the binary values of the color histograms correspond to the decision as to whether a pixel can or cannot belong to a given object segment. Model-based parameterizations of these color histograms are preferred for the efficient execution of the segmentation. In particular, they permit adaptation of the color histograms to temporally varying lighting conditions with acceptable expense.

Claims

1. A method for segmentation of digital color images with an electronic data processing system, comprising the steps of:

employing at least one binarized color histogram which describes the characteristic color properties of at least one imaged object for determining whether a pixel of a color image which is to be decomposed into object segments belongs to a specific object segment or not; and

checking whether there is assigned to said color value of said pixel a value of the binarized color histogram belonging to this object which indicates association of said pixel with said specific object segment.

2. The method according to claim 1 wherein the binarized color histogram is stored in the data processing system in the form of a table of binary values which assigns a first binary value to each color value which belongs to an object characterized by the color histogram, and assigns an opposite binary value to all other color values.

3. The method according to claim 1 in which values of the binarized color histogram are determined by applying a threshold operation to a normalized color histogram which is stored in the data processing system as a set of parameters of a parametric function.

4. The method according to claim 3 wherein the values of the parameters are adapted in a stepwise fashion to a temporally varying lighting.

5. The method according to claim 1 in which values of the binary color histogram are determined by applying a threshold operation to a normalized color histogram which is stored in the data processing system as a set of parametric functions.

6. The method according to claim 1 in which three-dimensional color histograms are used.

7. A method for segmentation of digital color images with an electronic data processing system, comprising the steps of:

employing at least one digitized color histogram which describes the characteristic color properties of at least one imaged object for determining whether a pixel of a color image which is to be decomposed into object segments belongs to a specific object segment or not;

determining the digitized color histogram by applying a threshold operation to a normalized color histogram which is stored in the data processing system as a set of parameters of a parametric function; and

checking whether there is assigned to said color value of said pixel a value of the digitized color histogram belonging to this object which indicates association of said pixel with said specific object segment.

8. The method according to claim 7 wherein the values of the parameters are adapted in a stepwise fashion to a temporally varying lighting.

9. A method for segmentation of digital color images with an electronic data processing system, comprising the steps of:

employing at least one digitized color histogram which describes the characteristic color properties of at least one imaged object for determining whether a pixel of a color image which is to be decomposed into object segments belongs to a specific object segment or not;

determining values of the digitized color histogram by applying a threshold operation to a normalized color histogram which is stored in the data processing system as a set of parametric functions; and

checking whether there is assigned to said color value of said pixel a value of the digitized color histogram belonging to this object which indicates association of said pixel with said specific object segment.