Files in this item

FilesDescriptionFormat

application/pdf

application/pdf3_Zhen Zuo_STAT578_Final_Report.pdf (742kB)
(no description provided)PDF

Description

Title:Camera Model Identification with Convolutional Neural Networks and Image Noise Pattern
Author(s):Zuo, Zhen
Subject(s):Deep Learning
Source Camera Identification
Noise Pattern
Abstract:Camera source detection has drawn a lot of attention in past decade. It enables us to solve a wide range of problems, from crime evidence identification to photo tampering detection. In this paper, some main methods people used in this area for past decade will be reviewed in the Introduction and Method sections. Also in the Method and Result sections, I compared and improved state-of-the-art approaches to solve camera model identification problem. In the latter part, I proposed a novel approaches based on Convolutional Neural Networks and Image Noise Pattern. Results on the dataset from Kaggle shows that to identify source camera, Convolutional Neural Networks can be applied to the pattern noise rather than directly to the original pictures to achieve at least similar or even better performance.
Issue Date:2018-05-12
Genre:Essay
Type:Text
URI:http://hdl.handle.net/2142/100123
Date Available in IDEALS:2018-07-02


This item appears in the following Collection(s)

Item Statistics