Sclera segmentation and recognition

Sclera recognition is emerging biometric technique in this era, and a lot is yet to be further studies. The sclera of the human eye comprises the exclusive retinal vasculature structures that make it very effective technique for individual identification. In this work, a speedy and vigorous sclera detection method is presented, and unique features for sclera-based recognition system are proposed. The finest adaptable data-set in terms of eye image attributes (UBIRIS V1) is chosen for detection, segmentation, and recognition of sclera. To detect sclera an unsupervised method based on pixel mapping and cauterized grayscale is detailed in the article. Thereafter, Local Binary Patterns (LBP) features and Gabor Wavelet Transform (GWT) of segmented sclera are further used for recognition. To find out the optimized feature set of LBP, Principal Component Analysis (PCA) is used. The classification was performed using the Multiclass Support Vector Machine (MSVM). The results of the classifier for the sclers recognition are obtainable according to average model sensitivity and accuracy. It is theoretically showed that the method presented in this paper is very consistent and accurate for the execution of the sclera recognition.