You are in:Home/Publications/Improved Hand Vein Pattern Recognition using Genetic Algorithm

Ass. Lect. mai maher abdelaziz abdelrasheed :: Publications:

Title:
Improved Hand Vein Pattern Recognition using Genetic Algorithm
Authors: Mai M.Zidan, Wael A.Mohamed, Ashraf S.Mohra, Khaled S.Ahmed,
Year: 2022
Keywords: Hand vein, Hough transformation, Genetic Algorithm (GA), K-nearest Neighbor (K-NN)
Journal: neuroquantology
Volume: Not Available
Issue: Not Available
Pages: Not Available
Publisher: neuroquantology
Local/International: Local
Paper Link:
Full paper mai maher abdelaziz abdelrasheed_20221208082150pmNQ99098 .pdf
Supplementary materials Not Available
Abstract:

Hand-vein recognition is known for its high accuracy and stability among biometric modalities. The best aspect to optimize is likely feature selection, it is a continuous challenge in human recognition systems. Feature selection aims to limit the number of features and eliminate redundant data and noise, resulting in a high recognition rate. The goal of our proposed method is to extract new properties from the hand vein, such as vein direction, length, and joined veins, which are thought to be unique to a person. Filtering techniques as well as enhancement and segmentation algorithms were applied to the collected data. The study was divided into two parts: the first employed the "Hough transformation," which is used to extract structural information such as vein lengths and angles, and the second used the Genetic Algorithm (GA). Instead of employing a mutation process, the Genetic Algorithm (GA) was altered to use a levy search. The algorithm has been shown to be an effective method of computing when the search space is judged to be highly dimensional. A classifier that uses the K-nearest neighbor (K-NN) algorithm is employed to detect all correct characteristics. Several experiments were conducted on the extracxted features, and the results revealed that this GA feature selection method can deliver excellent results with a small number of features. Finally, matching experiments were implemented for both parts, and the results obtained revealed that the second part yielded 100% accuracy compared to 99.5% reaching by traditional method.

Google ScholarAcdemia.eduResearch GateLinkedinFacebookTwitterGoogle PlusYoutubeWordpressInstagramMendeleyZoteroEvernoteORCIDScopus