Henry D. Knowles

Digital watermarking in wavelet domain with predistortion for authenticity verification and localization

In this paper, we present a blind fragile authentication algorithm by modifying a robust algorithm. The embedding process modifies the relative position of one wavelet coefficient from a vector of 3 coefficients. The introduced distortion of the watermarking system is reduced by a content dependent quantization parameter. This parameter refines the quantization step according to the magnitude of the coefficients in the vector. The smallest wavelet coefficients in the smooth areas of the image are pre-distorted to improve the performance and efficiency of the algorithm in these areas. This pre-distortion does not visually degrade the image as the introduced high frequency noise is evenly distributed over these areas. A dichotomous detector compares the extracted and embedded watermark on a bit by bit basis. This results in a high detection resolution, which can deliver information about the shape of the modified object. Embedding of the watermark with a larger redundancy increases the robustness of the system to additive white Gaussian noise attack. A weighted estimation then extracts the embedded watermark. This technique is fully described in the paper. Experimental results of this system embedded in the wavelet domain illustrate the performance and effectiveness compared with other reported fragile watermarking methods.

The Effect of Pixel-Level Fusion on Object Tracking in Multi-Sensor Surveillance Video

This paper investigates the impact of pixel-level fusion of videos from visible (VIZ) and infrared (IR) surveillance cameras on object tracking performance, as compared to tracking in single modality videos. Tracking has been accomplished by means of a particle filter which fuses a colour cue and the structural similarity measure (SSIM). The highest tracking accuracy has been obtained in IR sequences, whereas the VIZ video showed the worst tracking performance due to higher levels of clutter. However, metrics for fusion assessment clearly point towards the supremacy of the multiresolutional methods, especially Dual Tree-Complex Wavelet Transform method. Thus, a new, tracking-oriented metric is needed that is able to accurately assess how fusion affects the performance of the tracker.

Spatial digital watermark for MPEG-2 video authentication and tamper detection

The widespread adoption of digital video techniques has generated a requirement for authenticity verification in applications such as criminal evidence, insurance claims and commercial databases. This work addresses problems that arise from a spatial digital watermarking technique developed to detect frame reordering and dropping scenarios. It discusses the differences between mutual frame types at different bit-rates. Many papers consider detection after MPEG-2 decoding as a naïve approach. However, this approach does offer significant advantages for a slight increase of computation load. This paper also establishes a link between the detector performance and the sequence content. The uniqueness of this work is the comparison of the test results using 18 different standard MPEG test-sequences. The functionality of the algorithm is demonstrated with a simulated attack.

A Bayesian approach to attack characterisation using robust watermarks

In this paper we propose the use of a Bayesian framework to allow characterisation of image tampering from a library of attacks. We use the double watermarking strategy proposed in our previous work to derive sufficient information to drive the classifier. A non-parametric Bayesian classifier, trained on data derived from Monte Carlo simulations is used. In addition to classification, the effects of varying the input parameters are studied. The results obtained show that the non-parametric Bayesian classifier has a very low misclassification rate for this type of problem. Explanations as to the nature of the results, and some of the practical considerations, are given.

Compression-compatible digital watermark algorithm for authenticity verification and localization

In this paper, a new fragile watermarking system for authenticity verification is presented. This technique can detect and locate minor changes in a marked image. The method can be implemented in any domain. The embedding procedure modifies the representative value of a selected vector of coefficients according to the embedded watermark bit value. The mapping of the bits to the representative values and the formation of the orthogonal vectors are secured using a symmetric key system. The detector is a standalone system that does not need any prior knowledge about the original image or the embedded watermark. A tolerance bandwidth can be set to a minimum to reduce the level of false negative detector responses. Assuming there is a consistent pattern of tampering, an optimization algorithm has been designed to further reduce the false negative probability. By embedding the watermark in the same domain as is used for compression, the system can allow compression as an undetectable content preserving operation if the amount of quantization is known in advance. Experimental results of this system embedded in the DCT and Wavelet domain illustrate the performance and effectiveness compared with other reported fragile watermarking methods.

Towards tamper detection and classification with robust watermarks

In this paper, we present a novel wavelet-based double watermarking system for the detection and subsequent characterisation of the tampering of images. Most tamper detection techniques use fragile watermarks. However, our previous work showed that this type of watermark is frequently completely destroyed by compression, which in many cases is undesirable. In addition, it gives little or no information about the nature of the attack. We propose using two robust watermarks, one inserted at the embedder, the other at the detector. The second watermark is used as a reference with which the first watermark may be compared. This allows additional information not previously available about the nature of the attack to be obtained. The use of a robust strategy prevents the watermark being easily destroyed, but instead allows the system to continue to perform after significant attacks.

Efficient watermarking system with increased reliability for video authentication

The widespread adoption of digital video techniques has generated a requirement for authenticity verification in applications such as criminal evidence, insurance claims and commercial databases. This paper extends our previous work and improves the watermark estimation procedure of a spatial digital video watermarking system designed to detect and characterize time-base attacks. Most watermark extraction processes utilize the noise masking levels of the image. These levels change during transmission, especially when the host signal is compressed at low bit-rates. The blocking artifacts that are introduced by this process modify the noise masking levels and influence the ability to form a good estimate of the embedded watermark. This paper describes a novel filter procedure to eliminate these artifacts from the noise masking levels. The efficiency is compared with the standard MPEG-4 deblocking and deringing filters. Extracting the watermark from only the encoded Macroblocks, excluding the skipped Macroblocks improves the performance significantly without an increase in computational complexity. The functionality of this system within an MPEG-4 implementation is demonstrated with a receiver operating characteristic.

Bayesian approach to attack characterization using robust watermarks

In this paper we propose the use of a Bayesian framework to allow characterisation of image tampering from a library of attacks. We use the double watermarking strategy proposed in our previous work to derive sufficient information to drive the classifier. A non-parametric Bayesian classifier, trained on data derived from Monte Carlo simulations is used. In addition to classification, the effects of varying the input parameters are studied. The results obtained show that the non-parametric Bayesian classifier has a very low misclassification rate for this type of problem. Explanations as to the nature of the results, and some of the practical considerations, are given.

Attack characterisation of compressed images

We propose the use of robust watermarks to enable the characterisation of attacks even after lossy compression, such as JPEG and JPEG2000. A previously constructed Bayesian framework is used to allow characterisation of attacks from a predetermined library, and the double watermarking technique as earlier proposed by the authors is employed to generate the features used to drive the classifier. The results show that the developed techniques perform well for both types of compression.

An investigation of an MPEG-4 embedded spatial digital watermark developed for tamper detection and characterization

The widespread adoption of digital video techniques has generated a requirement for authenticity verification in applications such as criminal evidence, insurance claims and commercial databases. The paper investigates the performance, effectivity and efficiency of a spatial digital watermark designed for video authentication. The embedded watermark is estimated from the possible tampered frame with the use of a temporal filter and the motion information. The imperceptible digital stamp is robustly embedded prior to encoding (compression) and allows tamper characterisation of time-base attacks. The functionality of this system is evaluated within an MPEG-4 implementation.