College of Physics & Electronic Engineering Taizhou University Taizhou, China

---

Abstract—The redeye effect is typically formed in consumer photos taken with a built-in camera flash. The paper presents a new algorithm of the automatic removal of redeye from digital photos. The proposed algorithm can remove red-eye automatically without manual intervention. First, Detecting faces by using the AdaBoost algorithm. Second, the red-eyes are located by using operations of segmentation, morphology and geometric constraint. Finally, correcting the located red-eyes completely. The experimental results are satisfied with high correction rates, relatively low computational complexity and robustness.

Index Terms—Red-eye removal, AdaBoost, Face detection, Red-eye location.

[PDF]

Cite: Yueli Cui, Caiming Chen, Shiqing Zhang, Zhigang Chen, "A New Algorithm of Fully-automatic Red-eye Removal," Journal of Software vol. 8, no. 12, pp. 3042-3047, 2013.