Being one of the most active research topics in the computer vision field, automatic human gesture recognition is receiving increasing attention driven by its promising applications, ranging from surveillance and human monitoring, human-computer interface (HCI), and motion analysis, etc. Segmentation and recognition of human dynamic gestures from continuous video streams is considered to be a highly challenging task due to the spatio-temporal variation and endpoint localization issues. In this thesis, we propose a Motion Signature, which is a 3D spatio-temporal surface based upon the evolution of a contour over time, to reliably represent dynamic motion. A Gesture Model, is then constructed by a set of mean and variance images of Motion Signatures in a multi-scale manner, which not only is able to accommodate a wide range of spatio-temporal variation, but also has the advantage of requiring only a small amount of training data. Three approaches have been proposed to simultaneously segment and recognize gestures from continuous streams, which mainly differ in the way that the endpoints of gestures are located. While the first approach adopts an explicit multi-scale search strategy to find the endpoints of the gestures, the other two employ Dynamic Programming (DP) to handle this issue. All the three methods are rooted in the idea that segmentation and recognition are actually the two aspects of the same problem, and that the solution to either one of them will lead to the solution of the other. This is novel to most methods in the literature, which separate segmentation and recognition into two phases, and perform segmentation before recognition by looking into abrupt motion feature changes. The performance of the methods has been evaluated and compared on two types of gestures: two arms movement and a single hand movement. Experimental results have shown that all three methods achieved high recognition rates, ranging from 88% to 96% for upper body gestures, with the last one outperforming the other two. The single hand experiment also suggested that the proposed method has the potential to be applied to the application of continuous sign language recognition. / Thesis (Ph.D, Electrical & Computer Engineering) -- Queen's University, 2010-09-24 19:27:43.316

No description available.

The field of strategic marketing has long been identified as fruitful ground for gaining competitive advantage. Ever since the market segmentation concept was introduced in the late sixties, research interest steadily increased, covering issues as e.g. which fundamental segmentation strategy is most appropriate, in which ways can segments be identified or constructed, which algorithm provides optimal data-driven segmentation solutions, which number of segments should be constructed etc.. Interestingly, the issue of segment evaluation and choice has not been emphasised very strongly in the past, although this is of primary interest as soon as it comes to practical implementation. This article tries to fill this gap in an experimental manner: the consequences of different corporate segment choice strategies based on different segment evaluation criteria are investigated under different environmental conditions formalised in a complex artificial consumer market. The results indicate that complex decision models for segment choice do not turn out to be superior in general. Both mass marketers and firms concentrating on particular segments based on an a priori logic can be just as successful under "favourable" market conditions, the most influential condition being the available advertising budget. (author's abstract) / Series: Working Papers SFB "Adaptive Information Systems and Modelling in Economics and Management Science"

Two-photon calcium imaging (TPCI) is a functional neuroimaging technique that simultaneously reveals the function of small populations of cells as well as the structure of surrounding brain tissue. These unique properties cause TPCI to be increasingly popular for experimental basic neuroscience. Unfortunately, methodological development for data processing has not kept pace with experimental needs. I address this lack by developing and testing new methodology for several key tasks. Specifically, I address two primary analysis steps which are nearly universally required in early data processing: region of interest segmentation and motion correction. For each task I organize the sparse existing literature, clearly define the requirements of the problem, propose a solution, and evaluate it on experimental data. I develop MaSCS, an automated adaptable multi-class segmentation system that improves with use. I carefully define and describe the impact of motion artifacts on imaging data, and quantify the effects of standard and innovative motion correction approaches. Finally, I apply my work on segmentation and motion correction to explore one scientific target, namely discovering correlation-based cell clustering. I show that estimating such correlation-based clustering remains an open question, as it is highly sensitive to motion artifacts, even after motion correction techniques are applied. The contributions of this work include the organization of existing resources, methodological advances in segmentation, motion correction and clustering, and the development of prototype analysis software.

Image segmentation is an important step in the image analysis process. Current image segmentation techniques, however, require that the user tune several parameters in order to obtain maximum segmentation accuracy, a computationally inefficient approach, especially when a large number of images must be processed sequentially in real time. Another major challenge, particularly with medical image analysis, is the discrepancy between objective measures for assessing and guiding the segmentation process, on the one hand, and the subjective perception of the end users (e.g., clinicians), on the other. Hence, the setting and adjustment of parameters for medical image segmentation should be performed in a manner that incorporates user feedback. Despite the substantial number of techniques proposed in recent years, accurate segmentation of digital images remains a challenging task for automated computer algorithms. Approaches based on machine learning hold particular promise in this regard because, in many applications, including medical image analysis, frequent user intervention can be assumed as a means of correcting the results, thereby generating valuable feedback for algorithmic learning. This thesis presents an investigation of the use of evolving fuzzy systems for designing a method that overcomes the problems associated with medical image segmentation. An evolving fuzzy system can be trained using a set of invariant features, along with their optimum parameters, which act as a target for the system. Evolving fuzzy systems are also capable of adjusting parameters based on online updates of their rule base. This thesis proposes three different approaches that employ an evolving fuzzy system for the continual adjustment of the parameters of any medical image segmentation technique. The first proposed approach is based on evolving fuzzy image segmentation (EFIS). EFIS can adjust the parameters of existing segmentation methods and switch between them or fuse their results. The evolving rules have been applied for breast ultrasound images, with EFIS being used to adjust the parameters of three segmentation methods: global thresholding, region growing, and statistical region merging. The results for ten independent experiments for each of the three methods show average increases in accuracy of 5\%, 12\% and 9\% respectively. A comparison of the EFIS results with those obtained using five other thresholding methods revealed improvements. On the other hand, EFIS has some weak points, such as some fixed parameters and an inefficient feature calculation process. The second approach proposed as a means of overcoming the problems with EFIS is a new version of EFIS, called self-configuring EFIS (SC-EFIS). SC-EFIS uses the available data to estimate all of the parameters that are fixed in EFIS and has a feature selection process that selects suitable features based on current data. SC-EFIS was evaluated using the same three methods as for EFIS. The results show that SC-EFIS is competitive with EFIS but provides a higher level of automation. In the third approach, SC-EFIS is used to dynamically adjust more than one parameter, for example, three parameters of the normalized cut (N-cut) segmentation technique. This method, called multi-parametric SC-EFIS (MSC-EFIS), was applied to magnetic resonance images (MRIs) of the bladder and to breast ultrasound images. The results show the ability of MSC-EFIS to adjust multiple parameters. For ten independent experiments for each of the bladder and the breast images, this approach produced average accuracies that are 8\% and 16\% higher respectively, compared with their default values. The experimental results indicate that the proposed algorithms show significant promise in enhancing image segmentation, especially for medical applications.

No description available.

Thesis (PhD) - Institute for Social Research, Swinburne University of Technology, 2005. / Thesis submitted in fulfillment of the requirements of the degree of Doctor of Philosophy, Institute for Social Research, Swinburne University of Technology, 2005. Typescript. Bibliography: p. 188-207.

Texte remanié de: Th. Ph. D.--Cambridge (Mass.)--Massachusetts Institute of Technology, 1987.

Cette thèse est réalisée dans le cadre du projet ANR CONTINT DIADEMS sur l'indexation de documents ethnomusicologiques sonores. Les données que nous traitons sont fournies par les partenaires ethnomusicologues du projet et elles sont issues des archives du Musée de l'Homme de Paris. Les travaux effectués lors de cette thèse consistent à développer des méthodes permettant de faire une structuration automatique des documents musicaux et ethnomusicologiques basée sur les personnes. Cette thèse aborde le sujet encore inexploré à notre connaissance de la segmentation et du regroupement en chanteurs dans des enregistrements musicaux. Nous proposons un système complet pour ce sujet en s'inspirant des travaux réalisés en segmentation et regroupement en locuteurs. Ce système est conçu pour fonctionner aussi bien sur des enregistrements musicaux de type studio que sur des enregistrements musicaux réalisés dans des conditions terrain. Il permet, tout d'abord, de découper les zones de chant en des segments acoustiquement homogènes, i.e. en groupe de chanteur(s) afin d'avoir une segmentation en tours de chant. Ensuite, une phase de regroupement est effectuée afin de rassembler tous les segments chantés par un même groupe de chanteur(s) dans une seule classe. Notre première contribution est la définition de la notion de " tour de chant " et la proposition de règles d'annotation manuelle d'un enregistrement en des segments de tours de chant. La deuxième est la proposition d'une méthode de paramétrisation de la voix des chanteurs en implémentant une stratégie de sélection de bandes fréquentielles pertinentes basée sur la variance de celles-ci. La troisième est l'implémentation d'un algorithme de segmentation dynamique adapté à un contexte de chant en utilisant le Critère d'Information Bayésien (BIC). La quatrième est la proposition d'une méthode de Décision par Consolidation A Posteriori, nommée DCAP, pour pallier au problème de variabilité du paramètre de pénalité du BIC. En effet, comme le choix a priori d'une valeur optimale de ce paramètre n'est pas possible, nous effectuons un vote majoritaire sur plusieurs sorties de segmentations obtenues avec différentes valeurs de ce paramètre. Des gains d'environ 8% et 15% sont obtenus sur nos deux corpus avec cette méthode par rapport à une valeur standard du paramètre de pénalité. La cinquième est l'adaptation de la méthode DCAP pour la réalisation de l'étape de regroupement en chanteurs. / This work was done in the context of the ANR CONTINT DIADEMS project on indexing ethno-musicological audio recordings. The data that we are studying are provided by the Musée de l'Homme, Paris, within the context of this project. The work performed in this thesis consists of developing automatic structuring methods of musical and ethno-musicological documents based on the persons. This thesis touchs on an unexplored subject in our knowledge of the segmentation and clustering in singers of musical recordings. We propose a complete system in this subject that we called singer diarization by analogy with speaker diarization system on speech context. Indeed, this system is inspired from existing studies performed in speaker diarization and is designed to work on studio music recordings as well as on recordings with a variable sound quality (done outdoors). The first step of this system is the segmentation in singer turns which consists of segmenting musical recordings into segments "acoustically homogeneous" by singer group. The second step is the clustering which consists of labelling all segments produced by the same group of singers with a unique identifier. Our first contribution involved the definition of the term " singer turns " and the proposal of rules for manual annotation in singer turns segments. The second consisted in the proposal of a feature extraction method for the characterization of singer voices by implementing a method to select the frequency coefficients, which are the most relevant, based on the variance of these coefficients. The third is the implementation of a dynamic segmentation algorithm adapted to the singing context by using the Bayesian Information Criterion (BIC). The fourth is the proposal of a method, called DCAP, to take a posteriori decisions in order to avoid the variability problem of the BIC penalty parameter. Indeed, a priori choice of an optimal value for this parameter is not possible. This led us to perform a majority voting on a several segmentations obtained with different values of this parameter. A gain of about 8% and 15% is obtained on our two corpora with this method compared to the results found with a standard value of the penalty parameter. The fifth is the adaptation of our DCAP method in order to perform singer clustering step.

Ultrasound is one of the primary technologies used for clinical purposes. Ultrasound systems have favorable real-time capabilities, are fast and relatively inexpensive, portable and non-invasive. Recent interest in using ultrasound imaging for tongue motion has various applications in linguistic study, speech therapy as well as in foreign language education, where visual-feedback of tongue motion complements conventional audio feedback. Ultrasound images are known to be difficult to recognize. The anatomical structure in them, the rapidity of tongue movements, also missing segments in some frames and the limited frame rate of ultrasound systems have made automatic tongue contour extraction and tracking very challenging and especially hard for real-time applications. Traditional image processing-based approaches have many practical limitations in terms of automation, speed, and accuracy. Recent progress in deep convolutional neural networks has been successfully exploited in a variety of computer vision problems such as detection, classification, and segmentation. In the past few years, deep belief networks for tongue segmentation and convolutional neural networks for the classification of tongue motion have been proposed. However, none of these claim fully-automatic or real-time performance. U-Net is one of the most popular deep learning algorithms for image segmentation, and it is composed of several convolutions and deconvolution layers. In this thesis, we proposed a fully automatic system to extract tongue dorsum from ultrasound videos in real-time using a simplified version of U-Net, which we call sU-Net. Two databases from different machines were collected, and different training schemes were applied for testing the learning capability of the model. Our experiment on ultrasound video data demonstrates that the proposed method is very competitive compared with other methods in terms of performance and accuracy.