Synthetic Iris Dataset
Funded by the National Science Foundation (NSF) and the Center for Identification Technology Research (CITeR). To request the following datasets, please contact WVUBiometricData@mail.wvu.edu and indicate the specific dataset.Synthetic Iris Model Based:
The gallery of synthetic iris images are generated in five steps using a model based, anatomy based approach, with 40 controllable random parameters such as fiber size, pupil size, iris thickness, top layer thickness, fiber cluster degree, iris root blur range, the location of the collaret, the amplitude of the collaret, top layer transparency parameter, eye angle, eye size etc. The software coded with Matlab, generates 10000 classes (5000 subjects, left and right eye). Each class has 16 images, 1 good quality image, 15 degenerated images, with combination effects among noise, rotation, blur, motion blur, low contrast and specular reflection. For each image segmentation results ( unwrapped template, enhanced template and occlusion mask) are provided.
Performance is evaluated using a traditional Gabor filter based system. A comprehensive comparison of synthetic and real data is performed at three levels of processing: (1) image level, (2) texture level , and (3) decision level. A sensitivity analysis is performed to conclude on importance of various parameters involved in generating iris images.
Synthetic Iris Textured Based:
The synthetic irises are generated in two stages. In the first stage, a Markov Random Field model is used to generate a background texture representing the global iris appearance. In the next stage, a variety of iris features, viz., radial and concentric furrows, collaret and crypts, are generated and embedded in the texture field. 1000 classes with 7 samples per class.Examples of iris images:
When using synthetic iris images, please cite:
S. Shah and A. Ross, Generating Synthetic Irises By Feature Agglomeration, Proc.
of IEEE International Conference on Image Processing (ICIP), Atlanta, USA,
October 2006