Please use this identifier to cite or link to this item:
|Title:||Investigation of the trace transform applied to face recognition||Authors:||Thit Lwin Tun.||Keywords:||DRNTU::Engineering::Electrical and electronic engineering||Issue Date:||2006||Abstract:||In this dissertation, we investigate the performance of the face recognition rate using the Trace Transform algorithm. And also the effect of using different numbers of trace functionals, diametric functionals and circus functionals are studied in this dissertation. The ORL face database is used and it is split into two groups: the training set and the testing set. All images in these two sets are randomly selected and tested. We investigated the performance of the Trace Transform in three cases: case-1 uses only 7 trace functionals, case-2 uses 7 trace functionals, 1 diametric functional and 1 circus functional while case-3 uses 7 trace functionals, 3 diametric functionals and 3 circus functionals. After running for 10 times, case-1 gives the error rate 25.74 percent and case-2 shows error rate of 51.83 percent while case-3 has only 9.49 percent error rate. These results indicate that case-3 has better performance than the other two cases and so it can be seen obviously the importance of extracting more features in implementing the successful face recognition system.||Description:||64 p.||URI:||http://hdl.handle.net/10356/39124||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Theses|
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.