We present a system for finding and tracking a face and extract global and local animation parameters from a video sequence. The system uses an initial colour processing step for finding a rough estimate of the position, size, and inplane rotation of the face, followed by a refinement step drived by an active model. The latter step refines the previous estimate, and also extracts local animation parameters. The system is able to track the face and some facial features in near real-time, and can compress the result to a bitstream compliant to MPEG-4 face and body animation.

We address the 3D tracking of pose and animation of the human face in monocular image sequences using active appearance models. The classical appearance-based tracking suffers from two disadvantages: (i) the estimated out-of-plane motions are not very accurate, and (ii) the convergence of the optimization process to desired minima is not guaranteed. We aim at designing an efficient active appearance model, which is able to cope with the above disadvantages by retaining the strengths of feature-based and featureless tracking methodologies. For each frame, the adaptation is split into two consecutive stages. In the first stage, the 3D head pose is recovered using robust statistics and a measure of consistency with a statistical model of a face texture. In the second stage, the local motion associated with some facial features is recovered using the concept of the active appearance model search. Tracking experiments and method comparison demonstrate the robustness and out-performance of the developed framework.

We present a face and facial feature tracking system able to extract animation parameters describing the motion and articulation of a human face in real-time on consumer hardware. The system is based on a statistical model of face appearance and a search algorithm for adapting the model to an image. Speed and robustness is discussed, and the system evaluated in terms of accuracy.

This paper presents new methods for use of dense motion fields for motion compensation of interlaced video. The motion is estimated using previously decoded field-images. An initial motion compensated prediction is produced using the assumption that the motion is predictable in time. The motion estimation algorithm is phase-based and uses two or three field-images to achieve motion estimates with sub-pixel accuracy. To handle non-constant motion and the specific characteristics of the field-image to be coded, the initially predicted image is refined using forward motion compensation, based on block-matching. Tests show that this approach achieves higher PSNR than forward block-based motion estimation, when coding the residual with the same coder. The subjective performance is also better.

New strategies to improve neuron coupling to neuroelectronic implants are needed. In particular, tomaintain functional coupling between implant and neurons, foreign body response like encapsulation must meminimized. Apart from modifying materials to mitigate encapsulation it has been shown that with extremely thinstructures, encapsulation will be less pronounced. We here utilize wire electrochemical transistors (WECTs) usingconducting polymer coated fibers. Monofilaments down to 10 μm can be successfully coated and weaved intocomplex networks with built in logic functions, so called textile logic. Such systems can control signal patterns at alarge number of electrode terminals from a few addressing fibres. Not only is fibre size in the range where lessencapsulation is expected but textiles are known to make successful implants because of their soft and flexiblemechanical properties. Further, textile fabrication provides versatility and even three dimensional networks arepossible. Three possible architectures for neuroelectronic systems are discussed. WECTs are sensitive to dehydrationand materials for better durability or improved encapsulation is needed for stable performance in biologicalenvironments.

In this paper, we address the 3D tracking of pose and animation of the human face in monocular image sequences using deformable 3D models. The main contributions of this paper are as follows. First, we show how the robustness and stability of the Active Appearance Algorithm can be improved through the inclusion of a simple motion compensation based on feature correspondence. Second, we develop a new method able to adapt a deformable 3D model to a face in the input image. Central to this method is the decoupling of global head movements and local non-rigid deformations/animations. This decoupling is achieved by, first, estimating the global (rigid) motion using robust statistics and a statistical model for face texture, and then, adapting the 3D model to possible local animations using the concept of the Active Appearance Algorithm. This proposed method constitutes a significant step towards reliable model-based face trackers since the strengths of complementary tracking methodologies are combined.

Experiments evaluating the effectiveness of the methods are reported. Adaptation and tracking examples demonstrate the feasibility and robustness of the developed methods.

In this paper, we consider the potentialities of adapting a 3D deformable face model to video sequences. Two adaptation methods are proposed. The first method computes the adaptation using a locally exhaustive and directed search in the parameter space. The second method decouples the estimation of head and facial feature motion. It computes the 3D head pose by combining: (i) a robust feature-based pose estimator, and (ii) a global featureless criterion. The facial animation parameters are then estimated with a combined exhaustive and directed search. Tracking experiments and performance evaluation demonstrate the feasibility and usefulness of the developed methods. These experiments also show that the proposed methods can outperform the adaptation based on a directed continuous search.

The use of organic polymers for electronic functions is mainly motivated by the low-end applications, where low cost rather than advanced performance is a driving force. Materials and processing methods must allow for cheap production. Printing of electronics using inkjets¹ or classical printing methods has considerable potential to deliver this. Another technology that has been around for millennia is weaving using fibres. Integration of electronic functions within fabrics, with production methods fully compatible with textiles, is therefore of current interest, to enhance performance and extend functions of textiles². Standard polymer field-effect transistors require well defined insulator thickness and high voltage³, so they have limited suitability for electronic textiles. Here we report a novel approach through the construction of wire electrochemical transistor (WECT) devices, and show that textile monofilaments with 10–100 µm diameters can be coated with continuous thin films of the conducting polythiophene poly(3,4-ethylenedioxythiophene), and used to create micro-scale WECTs on single fibres. We also demonstrate inverters and multiplexers for digital logic. This opens an avenue for three-dimensional polymer micro-electronics, where large-scale circuits can be designed and integrated directly into the three-dimensional structure of woven fibres.

This work examines the possiblity to, with the computational power of today’s consumer hardware, employ techniques previously developed for 3D tracking of rigid objects, and use them for tracking of deformable objects. Our target objects are human faces in a video conversation pose, and our purpose is to create a deformable face tracker based on a head tracker operating in real-time on consumer hardware. We also investigate how to combine model-based and image based tracking in order to get precise tracking and avoid drift.

We consider the problem of choosing the rate for a source coder when transmitting lossy-compressed data in real-time over a channel with time-varying rate. The goal for the rate selection is to obtain a low average distortion while obeying a real-time constraint. We formulate the real-time constraint in terms of a limited buffer size. A few strategies for rate-control are suggested and evaluated for a Gilbert-Elliott type channel model. The results are also compared to a theoretical upper bound on performance for a rate-control algorithm working with constraints on buffer size.

State of the art and coming hyperspectral optical sensors generate large amounts of data and automatic analysis is necessary. One example is Automatic Target Recognition (ATR), frequently used in military applications and a coming technique for civilian surveillance applications. When sensors communicate in networks, the capacity of the communication channel defines the limit of data transferred without compression. Automated analysis may have different demands on data quality than a human observer, and thus standard compression methods may not be optimal. This paper presents results from testing how the performance of detection methods are affected by compressing input data with COTS coders. A standard video coder has been used to compress hyperspectral data. A video is a sequence of still images, a hybrid video coder use the correlation in time by doing block based motion compensated prediction between images. In principle only the differences are transmitted. This method of coding can be used on hyperspectral data if we consider one of the three dimensions as the time axis. Spectral anomaly detection is used as detection method on mine data. This method finds every pixel in the image that is abnormal, an anomaly compared to the surroundings. The purpose of anomaly detection is to identify objects (samples, pixels) that differ significantly from the background, without any a priori explicit knowledge about the signature of the sought-after targets. Thus the role of the anomaly detector is to identify “hot spots” on which subsequent analysis can be performed. We have used data from Imspec, a hyperspectral sensor. The hyperspectral image, or the spectral cube, consists of consecutive frames of spatial-spectral images. Each pixel contains a spectrum with 240 measure points. Hyperspectral sensor data was coded with hybrid coding using a variant of MPEG2. Only I- and P- frames was used. Every 10th frame was coded as P frame. 14 hyperspectral images was coded in 3 different directions using x, y, or z direction as time. 4 different quantization steps were used. Coding was done with and without initial quantization of data to 8 bbp. Results are presented from applying spectral anomaly detection on the coded data set.

An IP based set-top box (STB) is essentially a lightweight computer used to receive video over the Internet and convert it to analog or digital signals understood by the television. During this transformation from a digital image to an analog video signal many different types of distortions can occur. Some of these distortions will affect the image quality in a negative way. If these distortions could be measured they might be corrected and give the system a better image quality.

This thesis is a continuation of two previous theses where a custom hardware for sampling analog component video signals was created. A software used to communicatewith the sampling hardware and perform several different measurementson the samples collected has been created in this thesis.

The analog video signal quality measurement system has been compared to a similar commercial product and it was found that all except two measurement methods gave very good results. The remaining two measurement methods gave acceptable result. However the differences might be due to differences in implementation. The most important thing for the measurement system is to have consistency. If a system has consistency then any changes leading to worse videoquality can be found.

We introduce the concept of mobility-based communication in ad hoc networks, meaning that the packet transport is performed mainly by the nodes' movement. We outline a model for such networks, utilizing a stochastic model for the geographical location of the nodes. A test case is defined in which three strategies for packet forwarding are presented and evaluated.

Media such as movies and images are nowadays produced and distributed digitally. It is usually simple to make copies of digital content. Consequently illegal pirate copies can be duplicated and distributed in large quantities. One way to deter authorized content receivers from illegally redistributing the media is watermarking. If individual watermarks are contained in the digital media and a receiver is a pirate and redistributes it, the pirate at the same time distributes his identity. Thus a located pirate copy can be traced back to the pirate. The watermarked media should otherwise be indistinguishable from the original media content.

To distribute media content scalable transmission methods such as broadcast and multicast should be used. This way the distributor will only need to transmit the media once to reach all his authorized receivers. But since the same content is distributed to all receivers the requirement of individual watermarks seems to be contradictory.

In this thesis we will show how individually watermarked media content can be transmitted in a scalable way. Known methods will be reviewed and a new method will be presented. The new method is independent of what type of distribution that is used. A system with robust watermarks that are difficult to remove is described. Only small parts of the media content will be needed to identify the pirates. The method will only give a small data expansion compared to distribution of non-watermarked media.

We will also show how information theory tools can be used to expand the amount of data in the watermarks given a specific size of the media used for the watermarking. These tools can also be used to identify parts of the watermark that have been changed by deliberate deterioration of the watermarked media, made by pirates.

Watermarking embeds a signature in digital documents and can be used for discarding illegal copying. Fingerprinting means that every digital document have an individual signature. Multicast is a way to distribute the same digital object to many users without the need to sent one specific copy to each user. To combine distribution of fingerprinted videostreams with multicast might at first glance seem to be impossible. This paper will review some attempts to achieve this.

Watermarking embeds a signature in digital documents such as video and can be used for discarding illegal copying. Individual watermarking means that every customer receives an individual document that can be traced back to the customer if he is a pirate. Multicast is a way to distribute the same digital object to many users without the need to send one specific copy to each user. To combine distribution of individually watermarked documents with multicast might at first glance seem to be impossible. During the last years some methods have been developed to achieve that type of scalable distribution of individually watermarked documents. Some of these methods use special overlay networks to embed the signature and other use cryptography to produce the watermarks.

Point-of-interest detection is a way of reducing the amount of data that needs to be processed in a certain application and is widely used in 2D image analysis. In 2D image analysis, point-of-interest detection is usually related to extraction of local descriptors for object recognition, classification, registration or pose estimation. In analysis of range data however, some local descriptors have been published in the last decade or so, but most of them do not mention any kind of point-of-interest detection. We here show how to use an extended Harris detector on range data and discuss variants of the Harris measure. All described variants of the Harris detector for 3D should also be usable in medical image analysis, but we focus on the range data case. We do present a performance evaluation of the described variants of the Harris detector on range data.

We present an automatization of Barnsley's manual algorithm for the solution of the inverse problem of iterated function systems (IFSs). The problem is to retrieve the number of mappings and the parameters of an IFS from a digital binary image approximating the attractor induced by the IFS. Barnsley et al. described a way to manually solve the inverse problem by identifying the fragments, of which the collage is composed, and then computing the parameters of the mappings. The automatic algorithm searches through a finite set of points in the parameter space determining a set of affine mappings. The algorithm uses the collage theorem and the Hausdorff metric. The inverse problem of IFSs is related to image coding of binary images. If the number of mappings and the parameters of an IFS, with not too many mappings, could be obtained from a binary image, then this would give an efficient representation of the image. It is shown that the inverse problem solved by the automatic algorithm has a solution and some experiments show that the automatic algorithm is able to retrieve an IFS, including the number of mappings, from a digital binary image approximating the attractor induced by the IFS.