Year: 2019
Type: article
Abstract: In general, the efficiency and effectiveness of any detection, recognition, matching, and classification processes strongly depend on the size of the dataset and the features extracted [5,6].Thus, the quality of the extracted feature is one of the key issues [7].The drawback of feature selection lies in possible information loss in the process of selecting relevant features.Besides, it is difficult to determine the optimal number of significant features [7,8].The dimensionality of image size is reduced after choosing a set of features from the original feature set in feature extraction [6,9].Unlike feature rankingbased selection, feature extraction takes the relationship among the distinctive features into account.Hence this offsets the drawbacks in feature selection [7,9].Several methods have been proposed to extract the most beneficial features, however, they have yet to achieve good performances.The main difference between these methods of detecting forgery is in the choice of specific features and adoption to detect fraud, which plays a role in the success of particular algorithms, and their Research
DOI: 10.19101/ijacr.pid37
Authors: Nooraini Yusoff, Loai Alamro
Institutions: Al-Turath University, Universiti Utara Malaysia
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W2963122452