Red-eye detection method

• US 6,718,051 B1
• Filed: 10/16/2000
• Issued: 04/06/2004
• Est. Priority Date: 10/16/2000
• Status: Expired due to Fees

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Eschbach, Reiner
• Primary Examiner(s): Johns, Andrew W.
• Assistant Examiner(s): Alavi, Amir
• Application Number:
  US09/688,673
• Time in Patent Office: 1,268 Days
• Field of Search: 382/117, 382/162, 382/164, 382/165, 382/167, 382/218, 358/487, 358/500, 358/515, 358/518, 358/523, 358/530, 430/357, 430/359, 430/362, 430/364, 396/61, 396/62, 396/104, 396/121, 396/123, 396/165, 396/203
• US Class Current: 382/117
• CPC Class Codes: G06T 2207/30216 Redeye defect
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: G06V 40/193 Preprocessing; Feature extraction
• CPC Class Codes: H04N 1/56 Processing of colour picture signals H04N1/52 takes precedence
• CPC Class Codes: H04N 1/624 Red-eye correction control of camera based on recognised facial parts H04N23/611

Abstract

A red-eye detection method includes receiving digital image data that defines an image. The digital image data are processed to identify all regions of the image that include a specular reflection. Image regions deemed to include a specular reflection are processed further, according to conventional techniques, to determine the presence or absence of red-eye. Specular reflections are identified according to luminace-chrominance characteristics, geometric (e.g., size/shape) characteristics, and/or luminance gradient characteristics.

Claims

1. A digital image processing method for identifying potential red-eye regions of a digital image, said method comprising:

receiving a digital image defined by a plurality of pixels;

locating regions of interest within said digital image that include pixels that represent a specular reflection; and

,

identifying each region of interest as a potential red-eye region;

wherein said step of locating regions of interest in said digital image comprises;

examining luminance-chrominance data for each pixel of said digital image;

identifying a group of adjacent pixels of said digital image as a potential specular reflection if said group of adjacent pixels comprises pixels having luminance-chrominance characteristics of a specular reflection;

for each identified potential specular reflection, determining at least one geometric attribute of said potential specular reflection;

comparing the at least one geometric attribute of said potential specular reflection to a desired geometric attribute; and

,

identifying said potential specular reflection as a specular reflection if said at least one geometric attribute conforms to said desired geometric attribute.

2. The digital image processing method as set forth in claim 1, further comprising, before said step of examining luminance-chrominance data, converting each of said pixels of said digital image to a luminance-chrominance color space.

3. The digital image processing method of claim 1, wherein the at least one geometric attribute of the potential specular reflection is at least one of a size and shape of said potential specular reflection.

4. The digital image processing method as set forth in claim 1, further comprising, for each potential specular reflection:

determining a brightness gradient of said potential specular reflection;

comparing said brightness gradient of said potential specular reflection to a select brightness gradient threshold; and

,

identifying said potential specular reflection as a specular reflection if said at least one geometric attribute conforms to said desired geometric attribute and said brightness gradient is greater than said brightness gradient threshold.

5. A method for detecting red-eye in a digital image, said method comprising:

receiving digital image data defining a digital image;

identifying regions of interest in said digital image that include data that represent a specular reflection, wherein said step of identifying regions of interest comprises, for each region of an image;

identifying said region as including a specular reflection if said region includes a sub-region defined by at least one preset geometric attribute, a luminance above a select luminance threshold, and a luminance gradient above a select luminance gradient threshold;

examining the data in each of said regions of interest to detect red-eye.

6. The red-eye detection method as set forth in claim 5, wherein said at least one preset geometric attribute comprises at least one of a preset size and a preset shape.

7. A method of detecting red-eye in a digital image, said method comprising:

receiving a digital image;

identifying all regions of said digital image that include a specular reflection;

for each region of an image including a specular reflection, determining if said region includes red-eye based upon a color of said region adjacent said specular reflection;

wherein said step of identifying all regions of said digital image including a specular reflection comprises;

identifying all regions of said image including a potential specular reflection;

determining a geometric attribute of said potential specular reflection;

comparing said geometric attribute to a desired geometric attribute; and

,

identifying said region as a region including a specular reflection if said geometric attribute conforms to said desired geometric attribute.

8. The method as set forth in claim 7, wherein said step of determining a geometric attribute comprises determining at least one of a size attribute and a shape attribute, and wherein said comparing step comprises comparing said geometric attribute to a desired size attribute and shape attribute, respectively.

9. The method as set forth in claim 7, further comprising:

determining a luminance and a chrominance of said potential specular reflection;

comparing said luminance to a luminance threshold;

comparing said chrominance to a chrominance range;

identifying said region as a region including a specular reflection if said geometric attribute conforms to said desired geometric attribute, said luminance is greater than said luminance threshold, and said chrominance is within said chrominance range.

10. The method as set forth in claim 9, further comprising:

determining a luminance gradient of said potential specular reflection;

comparing said luminance gradient to a luminance gradient threshold; and

,

identifying said region as a region including a specular reflection if said geometric attribute conforms to said desired geometric attribute, said luminance is greater than said luminance threshold, said chrominance is within said chrominance range, and said luminance gradient is above said luminance gradient threshold.

11. The method as set forth in claim 10, wherein said step of determining a luminance gradient comprises a subtraction operation to determine a rate of change of luminance for said potential specular reflection.