It’s an all-too-common experience – the perfect photograph ruined by the demonic glow of the ‘red-eye’ effect. Now, a researcher at the University of Toronto has developed a method that can automatically remove those unsightly scarlet spots from digital images.
‘The technique will offer consumers a convenient, automatic tool for eliminating red eye in digital photographs,’ says Professor Konstantinos Plataniotis of the Edward S. Rogers Sr. Department of Electrical and Computer Engineering.
Red eye occurs when a camera flash reflects back from the retina of the eye, making the coloured portion of eyes appear red in photographs. The software algorithm Plataniotis has developed isolates the reddened areas of the eyes and automatically replaces the abnormally coloured region with natural eye colour.
The algorithm works in several steps. First, it detects the skin in the colour image with fast segmentation algorithms based on thresholding. Then, it processes the binary image depicting the skin-like regions using mathematical morphology. Next, it converts the color image into a gray-scale representation enhancing the color of the redeye region. Detection and segmentation of the image is then followed by colour replacement of the ‘red-eye’.
The technique, which could be used in cameras or labs that process digital images, could be available in about a year.
Beyond photography, the technology could be applied to medical imaging, virtual human-computer interfaces and to monitor fatigue by tracking the eyes of drivers, pilots or others in jobs that require quick reaction times.
The study co-authored by Professor Konstantinos appears in an upcoming issue of the journal Pattern Recognition Letters.