VicMan Software red eye correction technology

When the red-eye bounding rectangle is known, it is necessary to correct the unpleasant effect and restore the original look of the human eye. This task looks rather simple, but actually it is very hard to correct the problem in the way, where the eyes will look absolutely natural after the correction. We perform this in two steps:

1. Precise detection of the red-eye blob shape
2. Correction of the excessive redness of the red-eye area

Original image with red-eye regions set	Corrected image

The precise red-eye shape detection is performed by analyzing color balance over the eye pixels and loosely restricting the possible shapes of the red-eye blob. The pixels with reddish color in this detected area are labelled as those to be subjected to correction. Color correction is performed by adjusting color and brightness of these pixels into natural darker and less colorful human pupil color.

VicMan Software red eye detection technology

The unique technology of fast and accurate detection of red-eye effect on digital images developed by VicMan software incorporates latest achievemtns in the fields of Computer Vision and Image Analysis. A specially designed red-eye detection module scans the image is to detect possible instances of red-eye effect. Each image location that has close resemlbance with red-eye appearance is labelled as a potential red-eye candidate. Each red-eye candidate is assigned a 'confidence' level, that is proportinal to the probability that this image location exhibits red-eye effect.

Original image	Detected red eye regions

The post-processing stage collects red-eye candidates with high enough confidence level and estimates the center and bounding box of the detected red eye. This information is passed then to the red-eye correction module.

The red-eye detection module has been trained on carefully collected database of red-eye images. As our users submit us more photos - the database grows as does the intelligence of our red-eye detector.