EEL 851 - Biometrics
Spring 2006
Indian Institute of Technology Delhi, New Delhi

Assignment 1

User Authentication using Biometric Features

The objective of this assignment is to let you gain some hands on experience in the performance evaluation of a typical biometric system.

You are provided with a data file which has biometric feature sets from each of the 100 users. Each of the user has given his/her 10 image samples - 5 during enrollment and 5 six months after the enrollment. The feature vectors from each of these 1000 image samples are stored in the data file. The system has acquired 144 features from each of the biometric images and generated a representative feature vector.

You have to use the first five image samples as training features and remaining five as the one to report test results.

The following performance curves/scores on the test data using (i) Euclidean distance and (ii) cosine similarity as the matching distance is to be computed.

A. Genuine Distribution

B. Imposter Distribution

C. FAR Plot

D. FRR Plot

E. Receiver Operating Characteristics

F. Equal Error Rate and Decidability Index

 Which matching distance criteria in your opinion performs the best?

Points Distribution  [Maximum 50]

1. Implementation/Results from A and B ® Max 20 Points
2. Implementation/Results from C and D  ® Max 5 Points

3. Implementation/Results from E  ® Max 15 Points

4. Results from F  ® Max 10 Points

5. Clarity, Formatting and General Layout ® Up to -3 if not followed
6. Originality ® Both copier and originator gets ZERO
7. Late Policy ® 20% reduction, allows ONLY one day delay.

Submission

Please submit a tar file on-line containing the following:

• README file with
  • Your logins
  • Explanations about your implementation that may be useful for grader
• Your program

Your submission should be included in a tar file 'EntryID_Ass.tar' (e.g. 2004EE17091_Ass1.tar). This tar file can be uploaded using this link. You will be called for a short demo at a convenient time/date. Please make sure that you upload the assignment only once. In case of multiple submission ONLY first submission will be used for grading.

Deadline

The submission deadline for the assignment # 1 is 7 February 2006 [23:59 Hrs].

Questions

If you have any further questions relating to this assignment, you may contact the TA, Mr. Nihal Shrivastava [2004EET2251].