الفهرس | Only 14 pages are availabe for public view |
Abstract The work reported evlutes the performance of of feature extraction techniques such as Evolutionary Pursuit (EP), the holistetic Fourtier invariant features (HFIF), and Moment invariants. A new classifer depends on the genetic algorithm is introduced. The Evolutioary Purisit (EP) with PCA gives a poor recognition accury this limitation is alleviated by using DCT. All this techniques failed to recongnize rotaded images while \the proposed systems which depended on log polar transform (LPT) , the local Binary pattern (LBP 8, LBP 16 ,LBP 24 ),rotation invariant variance measures (VAR), and the joint distributions between LBP and VAR with adding synthetic images to the database improves the recognition accuracy widely. |