Tian-Tsong Ng

Tian-Tsong Ng 
 
Researcher
Computer Vision and Image Understanding Department
Institute for Infocomm Research
 
1 Fusionopolis Way, 
#21-01 Connexis,
Singapore 138632.
 
ttng@i2r.a-star.edu.sg
Phone: +65-6408.2517
Fax: +65-6776.1378
Background

Tian-Tsong Ng is a research staff in Institute for Infocomm Research. He received his M.Phil. in Signal Processing from Cambridge University in 2001. He received his Ph.D. in Electrical Engineering from Columbia University under the direction of Professor Shih-Fu Chang in 2007. His research focuses on digital image forensics, computer vision, and computational photography. He won the Microsoft Best Student Paper Award at ACM Multimedia Conference in 2005 and the John Wiley & Sons Best Paper Award at the first IEEE Workshop in Information Security and Forensics in 2009. He is a Commonwealth Scholar and an A*STAR Overseas Graduate Scholar.

Publication list

Projects

Camera Response Function (CRF) Signature For Digital Forensics

In our past work, we proposed that CRF attributes can be found on linear structures in an image and extracted using linear geometric invariants. In this work, we show additional properties on linear geometric invariants, propose a more robust way to select linear structures in an image, and provide a model-based method to extract CRF attributes from the linear structures. The proposed method extract reliable CRF signature extracted from a single image.

Multimedia Security @ I2R

In I2R, we have a series of research projects in the area of multimedia security at present and in the past. The slides provide a glimpse of the selected projects.

Radiometric Calibration Using Stratified Inverses

A projector-camera system is able to project a desired image on a non-flat and non-white surface after radiometric compensation, which can be achieved using the inverse light transport of the scene. This problem is known as view projection. A light transport matrix has a dimension of the camera pixel count times the projector pixel count. Therefore, computing the inverse light transport matrix is a daunting task. We propose an efficient way to perform view projection.

Single-image Recapture Image Detection

A photograph may be recaptured by a camera for various purposes. For example, Richard Prince a renowned artist created an art form from recapturing advertisement posters. An old photograph may be recaptured and digitzed for a restoration purpose. In a recent CNET article published in December 2008, it was demonstrated by a Vietnamese group that current face recognition systems deployed in the market for consumers are extremely vulnerable to playback attacks using recaptured images. The figure on the left consisting of both original and recaptured images shows that visual discrimination of the two image types is challenging.

Statistical and Geometric Methods for Passive-blind Image Forensics

Passive-blind image forensics (PBIF) refers to passive ways for evaluating image authenticity and detecting fake images. This dissertation proposes a physics-based approach for PBIF, with our definition of image authenticity derived from the image generative process comprising the 3D scene and the image acquisition device. We propose one statistical method and two geometric methods for capturing the image authenticity properties and addressing three separate problems in PBIF, i.e., detecting spliced images, distinguishing photographic images from photorealistic computer graphics, and estimating camera response function (CRF) from a single image.

Book Chapter (1.0MB) / Ph.D. Dissertation (7.1MB)

Using Geometry Invariants for Camera Response Function Estimation

We propose a method for estimating the camera response function (CRF) from a single-channel image by using geometry invariants. The idea is based on the observation that image gradient contains information about the camera response function. Geometry invariants are independent on the geometry of the locally planar regions in image irradiance. These planar regions in image irradiance is detected using the constraint equation (shown in the figure above). We also propose a generalized gamma curve model (GGCM) which fits well on the real-world CRF's.

Physics-based Photograph and Computer Graphics Classification

We address the issue of classifying photographic images (PIM) and photorealistic computer graphics (PRCG) images. The motivation is that PRCG can be used as image forgery. The problem is approached by analyzing the physical differences in the image formation process for the photographic images and PRCG. The differences are captured by a set of geometric features (from differential geometry, fractal geometry and local patches) in the linear Gaussian scale-space.

Dataset Online Demo

Image Splicing Detection

We address the photomontage detection problem, with an assumption of simple cut-and-paste (splicing) without any post-processing such as matting or blending. Bicoherence, a third-order moment spectrum, is used for detecting the splicing discontinuity. A functional texture-decomposition method is used to improve the detection performance. We also provide an analytical model for image splicing and gives theoretical results for explaining why bicoherence can be used for detecting image splicing. The theory is verified by experiments.

Dataset

Content Based Image Retrieval through Object Extraction and Querying

We propose a content based image retrieval system based on object extraction through image segmentation. A general and powerful multiscale segmentation algorithm automates the segmentation process, the output of which is assigned novel colour and texture descriptors which are both e±cient and effective. Query strategies consisting of a semi-automated and a fully automated mode are developed which are shown to produce good results. We then show the superiority of our approach over the global histogram approach which proves that the ability to access images at the level of objects is essential for CBIR.

PDF (0.2MB)

Kernel Mean-shift Algorithm for Data Clustering

We introduced a kernelized version of mean-shift algorithm. We showed that the algorithm shifts towards regions with denser data, adapts to the cluster distributions, and produce improved clustering performance.

M.Phil. Dissertation (2.7MB)

Group Members

Gao Xinting

Past Members

Bai Jiamin

Ramanpreet Singh Pahwa

Li Nannan

© 2009 Tian-Tsong Ng

Disclaimer : I am personally responsible for any content and opinion posted on this homepage. Institute for Infocomm Research is not responsible for any contents found on this homepage or any opinion expressed herein.

		Tian-Tsong Ng Researcher Computer Vision and Image Understanding Department Institute for Infocomm Research 1 Fusionopolis Way, #21-01 Connexis, Singapore 138632. ttng@i2r.a-star.edu.sg Phone: +65-6408.2517 Fax: +65-6776.1378

	Background
	Tian-Tsong Ng is a research staff in Institute for Infocomm Research. He received his M.Phil. in Signal Processing from Cambridge University in 2001. He received his Ph.D. in Electrical Engineering from Columbia University under the direction of Professor Shih-Fu Chang in 2007. His research focuses on digital image forensics, computer vision, and computational photography. He won the Microsoft Best Student Paper Award at ACM Multimedia Conference in 2005 and the John Wiley & Sons Best Paper Award at the first IEEE Workshop in Information Security and Forensics in 2009. He is a Commonwealth Scholar and an A*STAR Overseas Graduate Scholar. Publication list

	Projects
		Camera Response Function (CRF) Signature For Digital Forensics In our past work, we proposed that CRF attributes can be found on linear structures in an image and extracted using linear geometric invariants. In this work, we show additional properties on linear geometric invariants, propose a more robust way to select linear structures in an image, and provide a model-based method to extract CRF attributes from the linear structures. The proposed method extract reliable CRF signature extracted from a single image.
		Multimedia Security @ I2R In I2R, we have a series of research projects in the area of multimedia security at present and in the past. The slides provide a glimpse of the selected projects.

		Radiometric Calibration Using Stratified Inverses A projector-camera system is able to project a desired image on a non-flat and non-white surface after radiometric compensation, which can be achieved using the inverse light transport of the scene. This problem is known as view projection. A light transport matrix has a dimension of the camera pixel count times the projector pixel count. Therefore, computing the inverse light transport matrix is a daunting task. We propose an efficient way to perform view projection.

		Single-image Recapture Image Detection A photograph may be recaptured by a camera for various purposes. For example, Richard Prince a renowned artist created an art form from recapturing advertisement posters. An old photograph may be recaptured and digitzed for a restoration purpose. In a recent CNET article published in December 2008, it was demonstrated by a Vietnamese group that current face recognition systems deployed in the market for consumers are extremely vulnerable to playback attacks using recaptured images. The figure on the left consisting of both original and recaptured images shows that visual discrimination of the two image types is challenging.

		Statistical and Geometric Methods for Passive-blind Image Forensics Passive-blind image forensics (PBIF) refers to passive ways for evaluating image authenticity and detecting fake images. This dissertation proposes a physics-based approach for PBIF, with our definition of image authenticity derived from the image generative process comprising the 3D scene and the image acquisition device. We propose one statistical method and two geometric methods for capturing the image authenticity properties and addressing three separate problems in PBIF, i.e., detecting spliced images, distinguishing photographic images from photorealistic computer graphics, and estimating camera response function (CRF) from a single image. Book Chapter (1.0MB) / Ph.D. Dissertation (7.1MB)

		Using Geometry Invariants for Camera Response Function Estimation We propose a method for estimating the camera response function (CRF) from a single-channel image by using geometry invariants. The idea is based on the observation that image gradient contains information about the camera response function. Geometry invariants are independent on the geometry of the locally planar regions in image irradiance. These planar regions in image irradiance is detected using the constraint equation (shown in the figure above). We also propose a generalized gamma curve model (GGCM) which fits well on the real-world CRF's.

		Physics-based Photograph and Computer Graphics Classification We address the issue of classifying photographic images (PIM) and photorealistic computer graphics (PRCG) images. The motivation is that PRCG can be used as image forgery. The problem is approached by analyzing the physical differences in the image formation process for the photographic images and PRCG. The differences are captured by a set of geometric features (from differential geometry, fractal geometry and local patches) in the linear Gaussian scale-space. Dataset Online Demo

		Image Splicing Detection We address the photomontage detection problem, with an assumption of simple cut-and-paste (splicing) without any post-processing such as matting or blending. Bicoherence, a third-order moment spectrum, is used for detecting the splicing discontinuity. A functional texture-decomposition method is used to improve the detection performance. We also provide an analytical model for image splicing and gives theoretical results for explaining why bicoherence can be used for detecting image splicing. The theory is verified by experiments. Dataset

		Content Based Image Retrieval through Object Extraction and Querying We propose a content based image retrieval system based on object extraction through image segmentation. A general and powerful multiscale segmentation algorithm automates the segmentation process, the output of which is assigned novel colour and texture descriptors which are both e±cient and effective. Query strategies consisting of a semi-automated and a fully automated mode are developed which are shown to produce good results. We then show the superiority of our approach over the global histogram approach which proves that the ability to access images at the level of objects is essential for CBIR. PDF (0.2MB)

		Kernel Mean-shift Algorithm for Data Clustering We introduced a kernelized version of mean-shift algorithm. We showed that the algorithm shifts towards regions with denser data, adapts to the cluster distributions, and produce improved clustering performance. M.Phil. Dissertation (2.7MB)

	Group Members
	Gao Xinting

	Past Members
	Bai Jiamin Ramanpreet Singh Pahwa Li Nannan

© 2009 Tian-Tsong Ng Disclaimer : I am personally responsible for any content and opinion posted on this homepage. Institute for Infocomm Research is not responsible for any contents found on this homepage or any opinion expressed herein.