Local Orientation Based Feature Extraction


In recent years, iris recognition has received increasing attention due to its distinct characteristics. This

paper proposes a new approach to iris recognition based on local orientation description. Unlike the existing

methods for iris recognition, the proposed method characterizes the details of the iris from the viewpoint of

local orientation description. To effectively describe the randomness of the iris, a bank of Log-Gabor filters is

used to capture local orientation characteristics of the iris. The nearest center classifier is adopted for

classification. The proposed algorithm can achieve a high recognition rate of 100% on a set of 2,096 images.

Especially, when the probability of false match is 1/100,000, the false rejection rate is only 1.13%. Extensive

experiment results have demonstrated that orientation information is also an effective feature.



Junzhou Huang, Li Ma, and Yunhong Wang and Tieniu Tan,” Iris Recognition Based on Local Orientation Description”,

Asian Conference on Computer Vision, pp. 954-959, Korea, 2004.