DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Lee, Heung-Kyu | - |
dc.contributor.advisor | 이흥규 | - |
dc.contributor.author | Choi, Chang-Hee | - |
dc.contributor.author | 최창희 | - |
dc.date.accessioned | 2015-04-23T08:30:30Z | - |
dc.date.available | 2015-04-23T08:30:30Z | - |
dc.date.issued | 2013 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=566044&flag=dissertation | - |
dc.identifier.uri | http://hdl.handle.net/10203/197807 | - |
dc.description | 학위논문(박사) - 한국과학기술원 : 전산학과, 2013.8, [ ix, 81 p. ] | - |
dc.description.abstract | As contents has been converting from the analog method to the digital, digital forensic have been emerging as a hot issue. In digital image forensic, CFA pattern identification can be useful clues for source camera identification and forgery detection. In this thesis, we proposed the new system for detecting and estimating the image forgery based on CFA pattern trace. This system involves three main algorithms. First, we propose the method to identify the CFA pattern of a digital camera from a single digital image. Most of commercial cameras are equipped with CFA, and a digital image from these cameras has to be interpolated with a CFA interpolation algorithm. This process remains the CFA pattern trace, and it can be a key evidence for identifying the source digital camera. Many researches have been studied to detect CFA pattern trace. Furthermore, some researchers have studied the CFA pattern estimation. In this thesis, we designed the new approach to estimate CFA pattern of a digital image. We use basic principle of CFA interpolation which reconstructs blank pixel values from neighbor pixels. For each channel, we define neighbor pixel patterns which heavily influence a blank pixel as most heavyweight neighbor pixel patterns. We examine the intermediate value condition for all pixels in the test image and count it. The CFA pattern is determined by this counting information of three channels. Experimental results show that the proposed method achieves higher accuracy than previous methods. Since the proposed method do not need the huge size of filter, the experimental results with small area are also good. Second, we develop the novel algorithm to detect and estimate the color modification of digital images based on CFA pattern trace. Extensive studies have been carried out for detecting image forgery such as copy-move, re-sampling, blurring, and contrast enhancement. Although color modification is a common forgery technique, there is no reported forensic meth... | eng |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | image forensics | - |
dc.subject | 재 표본화 | - |
dc.subject | 색 변형 | - |
dc.subject | 컬러 필터 배열 | - |
dc.subject | 이미지 포렌직 | - |
dc.subject | hue | - |
dc.subject | color filter array | - |
dc.subject | color modification | - |
dc.subject | re-sampling | - |
dc.title | Digital image forgery detection and estimation system using CFA pattern trace of a digital camera | - |
dc.title.alternative | 디지털 카메라의 CFA 패턴 흔적을 이용한 디지털 이미지 위조 판별 시스템 | - |
dc.type | Thesis(Ph.D) | - |
dc.identifier.CNRN | 566044/325007 | - |
dc.description.department | 한국과학기술원 : 전산학과, | - |
dc.identifier.uid | 020105202 | - |
dc.contributor.localauthor | Lee, Heung-Kyu | - |
dc.contributor.localauthor | 이흥규 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.