TEXTURE ANALYSIS AND CLASSIFICATION BY DEEP ARCHITECTURES FOR PAPER FRAUD DETECTION

Date

2023-8-17

Author

Ekiz Emiroğlu, Ezgi

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

304
views

426
downloads

Cite This

This thesis aims to distinguish between fraudulent documents and original ones by analyzing the inherent textural structure present in the papers they are printed on. It is shown that the likelihood of two distinct sections from a paper sharing the same underlying textural structure is extremely low. The primary objective is to determine whether an object exists in the database or not (i.e., if it is original or fraudulent), which can be formalized as a Hypothesis Testing problem. To address this problem, a Siamese Network is utilized to extract discriminative features. By introducing a new weight term to the loss function of this base network, the identification of mismatched pairs is significantly improved compared to the classical methods, such as Gabor Filters and Local Binary Pattern. Subsequently, the learned embeddings from the base Siamese network are employed for the Hypothesis Testing problem by constructing a database with known objects and introducing unknown objects in the test phase. The problem can be viewed as comparing image of a unknown paper with those already encountered, using a suggested Meta Learning mechanism in the embedding space. Additionally, an end-to-end network is constructed to facilitate the objectives of both the Siamese Network and the Meta Learner. To demonstrate the effectiveness of proposed method with experiments, a dataset including paper sections is collected and subjected to a data augmentation schema. Additionally, experiments are conducted on a publicly available fabrics dataset. Systematic experiments reveal that the proposed method outperforms the baselines in terms of both accuracy and Type-II error (percentage of frauds predicted falsely as originals). The novel approach showcases improved performance in terms of Type-II error, effectively differentiating between genuine and fraudulent documents based on the textural structure analysis.

Subject Keywords

Texture Analysis, Image Matching, Siamese Networks, Meta Learning, Hypothesis Testing

URI

https://hdl.handle.net/11511/105954

Collections

Graduate School of Natural and Applied Sciences, Thesis