Automatic detection of red-eye artifacts in digital color photos

Red-eye is a flash-induced artifact that frequently occurs in photographs. This paper presents a novel method for unsupervised discovery of red-eye pixels in a digital image. Detection of face and/or skin regions in the image is not required by the algorithm. Analysis is performed primarily in the hue-saturation-value (HSV) color space. A flash mask, which defines the regions where red-eye regions may be present, is first extracted from the brightness component. Subsequent processing on the other color components prunes the number of candidate regions that may correspond to red-eye. Algorithm performance is demonstrated on content from multiple digital cameras with different resolutions.