Departamento de Informática da Universidade da Beira Interior
SOCIA Lab. – Soft Computing and Image Analysis Group
Department of Computer Science, University of Beira Interior, 6201-001 Covilhã, Portugal
1. Overview. In order to participate in the NICE:II contest, a registration form and an application executable should be submitted.
1.1. The executable should receive (through command-line parameters) the path of 4 images (2 iris images + 2 corresponding segmentation maps) and output a text file (which name and path is received in the 5th parameter) that contains a numeric value corresponding to the dissimilarity between these iris images.
1.2. The outputted dissimilarity value should be a metric. Let I be a set of iris images. A metric on I is a function d:I x I --> R that satisfies the following conditions:
1.2.1. d(I1, I1)=0
1.2.2. d(I1, I2)=0 -->I1=I2
1.2.3. d(I1, I2) + d(I1, I3) >= d(I2, I3)
1.3. Example of the application functioning, through command-line:
> app I1.jpg I1.bmp I2.jpg I2.bmp I1_I2.txt
An overview of the task demanded to the NICE:II participants is given in figure 1:
Figure 1: NICE.II fundamental task.
1.4. The application executable can be written in any programming language (Matlab inclusively) and should run in one of the operating systems:”Windows XP, Service Pack 2” or ”Fedora Core 6”.
1.5. There will be no internet access during the NICE:II evaluation. Thus, the application executable will need to be installed and executed without access to the internet.
2. Registration. Each NICE:II participant will receive a username. This username should be used as name of the submitted executable.
2.1. Each participant is allowed to submit one single algorithm and executable.
2.2. NICE:II participation agreement. The application form must be filled and sent to the contest email address.
3. Evaluation. The NICE:II contest will be evaluated by a Java framework built within the SOCIA Lab and within the UBIRIS.v2 data set.
3.1. The evaluation framework will be available to NICE:II participants, in order to facilitate the training and tuning of their iris classification methods.
3.2. Together with the evaluation framework, a data set of noisy iris images (portion of the UBIRIS.v2 database) will be given, with have close characteristics to the images used in the evaluation stage. Additionally, the set of the correspondent and manually classified “.bmp” images will also be given.
3.2.1. The image format of the provided data set of input images is “.tiff”.
3.2.2. The binary maps that give the segmentation of the iris images have the same name and “.bmp” format.
DI-UBI Bloco VI Rua Marquês de Ávila e Bolama P- 6201-001 Covilhã PORTUGAL