Shih-Wei Sun

Media hash-dependent image watermarking resilient against both geometric attacks and estimation attacks based on false positive-oriented detection

The major disadvantage of existing watermarking methods is their limited resistance to extensive geometric attacks. In addition, we have found that the weakness of multiple watermark embedding methods that were initially designed to resist geometric attacks is their inability to withstand the watermark-estimation attacks (WEAs), leading to reduce resistance to geometric attacks. In view of these facts, this paper proposes a robust image watermarking scheme that can withstand geometric distortions and WEAs simultaneously. Our scheme is mainly composed of three components: 1) robust mesh generation and mesh-based watermarking to resist geometric distortions; 2) construction of media hash-based content-dependent watermark to resist WEAs; and 3) a mechanism of false positive-oriented watermark detection, which can be used to determine the existence of a watermark so as to achieve a tradeoff between correct detection and false detection. Furthermore, extensive experimental results obtained using the standard benchmark (i.e., Stirmark) and WEAs, and comparisons with relevant watermarking methods confirm the excellent performance of our method in improving robustness. To our knowledge, such a thorough evaluation has not been reported in the literature before

Robust mesh-based hashing for copy detection and tracing of images

Due to the desired non-invasive property, non-data hiding (called media hashing here) is considered to be an alternative to achieve many applications previously accomplished with watermarking. Recently, media hashing techniques for content identification have been gradually emerging. However, none of them are really resistant against geometrical attacks. In this paper, our aim is to propose a geometry-invariant image hashing scheme, which can be employed for content copy detection and tracing. Our system is mainly composed of three components: (i) robust mesh extraction; (iii) mesh-based robust hash extraction; and (iii) hash matching for similarity measurement. Exhaustive experimental results obtained from benchmark attacks have confirmed the performance of the proposed method

Moving foreground object detection via robust SIFT trajectories

In this paper, we present an automatic foreground object detection method for videos captured by freely moving cameras. While we focus on extracting a single foreground object of interest throughout a video sequence, our approach does not require any training data nor the interaction by the users. Based on the SIFT correspondence across video frames, we construct robust SIFT trajectories in terms of the calculated foreground feature point probability. Our foreground feature point probability is able to determine candidate foreground feature points in each frame, without the need of user interaction such as parameter or threshold tuning. Furthermore, we propose a probabilistic consensus foreground object template (CFOT), which is directly applied to the input video for moving object detection via template matching. Our CFOT can be used to detect the foreground object in videos captured by a fast moving camera, even if the contrast between the foreground and background regions is low. Moreover, our proposed method can be generalized to foreground object detection in dynamic backgrounds, and is robust to viewpoint changes across video frames. The contribution of this paper is trifold: (1) we provide a robust decision process to detect the foreground object of interest in videos with contrast and viewpoint variations; (2) our proposed method builds longer SIFT trajectories, and this is shown to be robust and effective for object detection tasks; and (3) the construction of our CFOT is not sensitive to the initial estimation of the foreground region of interest, while its use can achieve excellent foreground object detection results on real-world video data.

Automatic annotation of Web videos

Most Web videos are captured in uncontrolled environments (e.g. videos captured by freely-moving cameras with low resolution); this makes automatic video annotation very difficult. To address this problem, we present a robust moving foreground object detection method followed by the integration of features collected from heterogeneous domains. We advance SIFT feature matching and present a probabilistic framework to construct consensus foreground object templates (CFOT). The CFOT can detect moving foreground objects of interest across video frames, and this allows us to extract visual features from foreground regions of interest. Together with the use of audio features, we are able to improve resulting annotation accuracy. We conduct experiments and achieve promising results on a Web video dataset collected from YouTube.

An online people counting system for electronic advertising machines

This paper presents a novel people counting system for an environment in which a stationary camera can count the number of people watching a TV-wall advertisement or an electronic billboard without counting the repetitions in video streams in real time. The people actually watching an advertisement are identified via frontal face detection techniques. To count the number of people precisely, a complementary set of features is extracted from the torso of a human subject, as that part of the body contains relatively richer information than the face. In addition, for conducting robust people recognition, an online classifier trained by Fishers Linear Discriminant (FLD) strategy is developed. Our experiment results demonstrate the efficacy of the proposed system for the people counting task.

Robust hash-based image watermarking with resistance to geometric distortions and watermark-estimation attack

Digital watermarking provides a feasible way for copyright protection of multimedia. The major disadvantage of the existing methods is their limited resistance to both extensive geometric distortions and watermark-estimation attack (WEA). In view of this fact, this paper aims to propose a robust image watermarking scheme that can withstand geometric distortions and WEA simultaneously. Our scheme is mainly composed of two components: (i) mesh generation and embedding for resisting geometric distortions; and (ii) construction of hash-based content-dependent watermark (CDW) for resisting WEA. Extensive experimental results obtained from standard benchmark confirm the ability of our method in improving robustness.

Video watermarking synchronization based on profile statistics

A novel temporal synchronization method for video watermarking by matching the profile statistics. The profile statistics, represented by the characteristic parameters such as position mean and variance in the x-and y-directions, of a frame in a video sequence can easily be calculated and sent as side information to the receiver. At the receiving end, temporal attacks such as transposition, dropping, and insertion can be detected by comparing side information and characteristic parameters calculated from the received video. The simulation results show that the proposed method can successfully re-synchronize the attacked video back to the original format with accuracy from 72.41% to 98.15% for various video sequences based on single frame matching. After the voting process, the GOP detection accuracy can be improved to the range of from 96.30% to 100%.

Whac-a-mole: A head detection scheme by estimating the 3D envelope from depth image

Interactive multimedia display has attracted great attention in recent years. However, most of the existing systems lack the user-aware capability, i.e. blind to the viewers height and spatial location in a real-world 3D space, and often fail to provide a natural interaction. Therefore, in this work, we propose a probabilistic framework for detecting the viewer, (i.e. human heads) in depth images from a birds eye view camera. In comparison to the state-of-the-art approaches, the experimental results demonstrated that the proposed framework can provide higher detection rate but also real-time execution. Even a large number of people are walking or standing together shoulder-by-shoulder, the proposed probabilistic head detection scheme is still able to give promising people detection capability.

Biometric Template Protection: A Key-Mixed Template Approach

This paper presents key-mixed template (KMT), which mixes a users template with a secret key to generate another form of template, to prevent the biometric template stored in the database from back end attack, snooping, and tampering attack.

People tracking in an environment with multiple depth cameras

The hand-gesture-triggered calibration can build the geometry in a region of interest.The proposed interleaving-based skeleton obtaining can track more people.The proposed pairwise trajectory matching scheme manages occlusion situations. This paper proposes a pairwise trajectory matching scheme from multiple cameras for people tracking, handling the mistracking situations caused by occlusion events occurred in one of the cameras. In a multiple cameras environment, a geometric calibration process is necessary for the co-plane of the overlapping field of views from different cameras as the initial step. Once the geometry is calibrated, according to the 2D positions of the analyzed foot joints from the depth cameras. Homography transformation is applied to project the detected foot points from different views into a synergistic virtual birds eye view for people tracking. At the virtual birds eye view, the people tracking results from each of the cameras based on Kalman filter are fused according to the proposed pairwise trajectory matching scheme. The contribution of this paper is trifold: (1) The proposed hand-gesture-triggered calibration process with temporally synchronization capability can effectively build and calibrate the geometry in a region of interest. (2) The proposed interleaving-based skeleton obtaining and moving average based valid skeleton determination can extend the skeleton tracking capability to track more people. (3) The proposed pairwise trajectory matching scheme effectively manages occlusion situations happened in one of the depth cameras. In addition, in the extensive experimental results, the proposed method can track up to six simultaneously freely moving persons in the field of view, with affordable complexity for real-time applications. Furthermore, the infrared-based depth cameras track people satisfactorily from bright to extremely dark environments.