Exploration of scale invariant feature transform (sift) on paper fingerprinting using flatbed scanner

• Main Author: Khew, Ka Eian
• Other Authors: Qian Kemao
• Format: Final Year Project
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/74071
• Institution: Nanyang Technological University

Document verification is widely used in many fields such as academic certification, custom inspection (passport and identification card), ticketing, painting, and banking. It is very important to verify the facticity of the document. This is to avoid the abuse of document for identification. Therefore, a lot of effort has been done for anti-counterfeiting, yet this issue is still very serious as most of them used in practical has its own limitations such as the cost and time complexity to do the verification. To overcome those limitations, a novel idea was proposed by Clarkson et al. (2009) [1] which suggested that surface texture of paper can be scanned by flatbed scanner is usable for the verification of paper document. This process is called “Paper Fingerprinting”. The idea of this Final Year Project (FYP) was mainly inspired by the framework proposed by Clarkson et al in 2009 and its earlier implementation was done by previous student Li Zixuan and continued enhancement by Du Qiu. In this project, the suggested model was further modified to adapt with Scale invariant feature Transform (SIFT) which is a common technique for feature detection and extraction. This modification is done with the goal to improve the usability and robustness of the application as SIFT is a reliable feature extraction algorithm. The main intention of this project is to enhance the features of previous version of application. There are some features needed for improvement as they are making the paper fingerprinting process taking longer such as manual image cropping and multiple scanning of image in different orientations in order to extract three-dimensional feature vector. However, the process can be simplified by adapting SIFT into the key feature extraction process as SIFT is well known for its scale and rotation invariant properties. An addition for the previous prototype, this projects also tested for printed paper and studied the effects of printing on paper fingerprinting process.