SSIM As Validation Technique On Normalization Segmented Iris

As one of biomedical technique to help diagnose person systemic health based on pattern colors and other characteristic on iris, Normalization has important role to provide data in the form of images that can make it easier for users to be able to better observe the patterns that exist in the iris. However, it should be noted that there is no guarantee that the exact iris segmentation results will have similarities with the other resulting image data. To overcome this issue, the approach through SSIM (Structural Similarity) is used to see the level of similarity possessed by each segmented iris image which is taken from the same person as the iris image data validation method. In the test, CASIA-Iris-Interval V3 was used as a dataset of iris images taken using a NIR (near infrared) Camera with 249 subjects, 2639 images (1332 left eye, 1307 right eye). Meanwhile, the segmentation process is carried out using a hybrid method. Based on the results obtained, the accuracy of iris detection reached 96.1 % in the right eye and 96.3% in the left eye. While the average level of similarity between the results of segmented iris normalization only reached 77.9755 % on right eye and 77.8547% on left eye. This illustrates that even though the object of the image taken is from the same individual, the condition of the image capture and the eye condition of that individual affect the results of the normalization carried out.