Optimization of Gabor Wavelets for Face Recognition

The article describes researches of a method of person recognition by face image based on Gabor wavelets. Scales of Gabor functions are determined at which the maximal percent of recognition for search of a person in a database and minimal percent of mistakes due to false alarm errors when solving an access control task is achieved. The carried out researches have shown a possibility of improvement of recognition system work parameters in the specified two modes when the volume of used data is reduced.

Annotated record of the detailed examination of Mn deposits from R/V Velero IV Cruise 404 stations

The geographical, physical and biological aspects of the submarine canyons of the continental shelf off the coast of southern California have been described in earlier parts of this volume. Isopods were collected in 10 of the 15 canyons. Many benthic species were obtained since the specimens were obtained with a Campbell grab bottom sampler operated from the Hancock Foundation research vessel Velero IV.

Soundscape Ecology of Hawaiian Spinner Dolphin Resting Bays

Sound is a key sensory modality for Hawaiian spinner dolphins. Like many other marine animals, these dolphins rely on sound and their acoustic environment for many aspects of their daily lives, making it is essential to understand soundscape in areas that are critical to their survival. Hawaiian spinner dolphins rest during the day in shallow coastal areas and forage offshore at night. In my dissertation I focus on the soundscape of the bays where Hawaiian spinner dolphins rest taking a soundscape ecology approach. I primarily relied on passive acoustic monitoring using four DSG-Ocean acoustic loggers in four Hawaiian spinner dolphin resting bays on the Kona Coast of Hawai‛i Island. 30-second recordings were made every four minutes in each of the bays for 20 to 27 months between January 8, 2011 and March 30, 2013. I also utilized concomitant vessel-based visual surveys in the four bays to provide context for these recordings. In my first chapter I used the contributions of the dolphins to the soundscape to monitor presence in the bays and found the degree of presence varied greatly from less than 40% to nearly 90% of days monitored with dolphins present. Having established these bays as important to the animals, in my second chapter I explored the many components of their resting bay soundscape and evaluated the influence of natural and human events on the soundscape. I characterized the overall soundscape in each of the four bays, used the tsunami event of March 2011 to approximate a natural soundscape and identified all loud daytime outliers. Overall, sound levels were consistently louder at night and quieter during the daytime due to the sounds from snapping shrimp. In fact, peak Hawaiian spinner dolphin resting time co-occurs with the quietest part of the day. However, I also found that humans drastically alter this daytime soundscape with sound from offshore aquaculture, vessel sound and military mid-frequency active sonar. During one recorded mid-frequency active sonar event in August 2011, sound pressure levels in the 3.15 kHz 1/3rd-octave band were as high as 45.8 dB above median ambient noise levels. Human activity both inside (vessels) and outside (sonar and aquaculture) the bays significantly altered the resting bay soundscape. Inside the bays there are high levels of human activity including vessel-based tourism directly targeting the dolphins. The interactions between humans and dolphins in their resting bays are of concern; therefore, my third chapter aimed to assess the acoustic response of the dolphins to human activity. Using days where acoustic recordings overlapped with visual surveys I found the greatest response in a bay with dolphin-centric activities, not in the bay with the most vessel activity, indicating that it is not the magnitude that elicits a response but the focus of the activity. In my fourth chapter I summarize the key results from my first three chapters to illustrate the power of multiple site design to prioritize action to protect Hawaiian spinner dolphins in their resting bays, a chapter I hope will be useful for managers should they take further action to protect the dolphins.

Development and Optimization of Four-dimensional Magnetic Resonance Imaging (4D-MRI) for Radiation Therapy

A tenet of modern radiotherapy (RT) is to identify the treatment target accurately, following which the high-dose treatment volume may be expanded into the surrounding tissues in order to create the clinical and planning target volumes. Respiratory motion can induce errors in target volume delineation and dose delivery in radiation therapy for thoracic and abdominal cancers. Historically, radiotherapy treatment planning in the thoracic and abdominal regions has used 2D or 3D images acquired under uncoached free-breathing conditions, irrespective of whether the target tumor is moving or not. Once the gross target volume has been delineated, standard margins are commonly added in order to account for motion. However, the generic margins do not usually take the target motion trajectory into consideration. That may lead to under- or over-estimate motion with subsequent risk of missing the target during treatment or irradiating excessive normal tissue. That introduces systematic errors into treatment planning and delivery. In clinical practice, four-dimensional (4D) imaging has been popular in For RT motion management. It provides temporal information about tumor and organ at risk motion, and it permits patient-specific treatment planning. The most common contemporary imaging technique for identifying tumor motion is 4D computed tomography (4D-CT). However, CT has poor soft tissue contrast and it induce ionizing radiation hazard. In the last decade, 4D magnetic resonance imaging (4D-MRI) has become an emerging tool to image respiratory motion, especially in the abdomen, because of the superior soft-tissue contrast. Recently, several 4D-MRI techniques have been proposed, including prospective and retrospective approaches. Nevertheless, 4D-MRI techniques are faced with several challenges: 1) suboptimal and inconsistent tumor contrast with large inter-patient variation; 2) relatively low temporal-spatial resolution; 3) it lacks a reliable respiratory surrogate. In this research work, novel 4D-MRI techniques applying MRI weightings that was not used in existing 4D-MRI techniques, including T2/T1-weighted, T2-weighted and Diffusion-weighted MRI were investigated. A result-driven phase retrospective sorting method was proposed, and it was applied to image space as well as k-space of MR imaging. Novel image-based respiratory surrogates were developed, improved and evaluated.

Visibilidad y percepción en la construcción de un paisaje Prehistórico. El caso de Calviá (Mallorca, Islas Baleares)

El presente trabajo se centra en el estudio del papel que juega el control visual del espacio en las prácticas sociales de las comunidades prehistóricas. Este trabajo se articula a partir de un estudio de caso, el término municipal de Calviá, situado en el sureste de la isla de Mallorca, para analizar las diferentes formas de monumentalidad arquitectónica y cómo estas se constituyen cómo un punto de referencia social dentro del paisaje. Partiendo de una amplia horquilla temporal, que abarcaría el Bronce Naviforme (1550-850 AC), el período Talayótico (850-550 AC) y el Postalayótico (550-123 AC), se propone analizar los cambios y pervivencias en la construcción del paisaje, a través de estrategias de visibilidad, percepción y movimiento alrededor de los monumentos arquitectónicos. A través de la perspectiva de la Arqueología del Paisaje y mediante el uso de Sistemas de Información Geográfica (SIG) se propone un análisis de tendencias a largo plazo en la configuración social de un paisaje.

Mem Tri: Memory Forensics Triage Tool

This work explores the development of MemTri. A memory forensics triage tool that can assess the likelihood of criminal activity in a memory image, based on evidence data artefacts generated by several applications. Fictitious illegal suspect activity scenarios were performed on virtual machines to generate 60 test memory images for input into MemTri. Four categories of applications (i.e. Internet Browsers, Instant Messengers, FTP Client and Document Processors) are examined for data artefacts located through the use of regular expressions. These identified data artefacts are then analysed using a Bayesian Network, to assess the likelihood that a seized memory image contained evidence of illegal activity. Currently, MemTri is under development and this paper introduces only the basic concept as well as the components that the application is built on. A complete description of MemTri coupled with extensive experimental results is expected to be published in the first semester of 2017.

Dynamically balanced and plausible trajectory planning for human-like characters

International audience

This Is a Gorgeous-looking Movie”: Multimodal Interplay between Commentary and Subject Materials in Movie Review Entertainment

The present thesis is a study of movie review entertainment (MRE) which is a contemporary Internet-based genre of texts. MRE are movie reviews in video form which are published online, usually as episodes of an MRE web show. Characteristic to MRE is combining humor and honest opinions in varying degrees as well as the use of subject materials, i.e. clips of the movies, as a part of the review. The study approached MRE from a linguistic perspective aiming to discover 1) whether MRE is primarily text- or image-based and what the primary functions of the modes are, 2) how a reviewer linguistically combines subject footage to her/his commentary?, 3) whether there is any internal variation in MRE regarding the aforementioned questions, and 4) how suitable the selected models and theories are in the analysis of this type of contemporary multimodal data. To answer the aforementioned questions, the multimodal system of image—text relations by Martinec and Salway (2005) in combination with categories of cohesion by Halliday and Hasan (1976) were applied to four full MRE videos which were transcribed in their entirety for the study. The primary data represent varying types of MRE: a current movie review, an analytic essay, a riff review, and a humorous essay. The results demonstrated that image vs. text prioritization can vary between reviews and also within a review. The current movie review and the two essays were primarily commentary-focused whereas the riff review was significantly more dependent on the use of imagery as the clips are a major source of humor which is a prominent value in that type of a review. In addition to humor, clips are used to exemplify the commentary. A reviewer also relates new information to the imagery as well as uses two modes to present the information in a review. Linguistically, the most frequent case was that the reviewer names participants and processes lexically in the commentary. Grammatical relations (reference items such as pronouns and adverbs and conjunctive items in the riff review) were also encountered. There was internal variation to a considerable degree. The methods chosen were deemed appropriate to answer the research questions. Further study could go beyond linguistics to include, for instance, genre and media studies.

A comparative study of navigation meshes

International audience

Context Models for Understanding Images and Videos

A computer vision system that has to interact in natural language needs to understand the visual appearance of interactions between objects along with the appearance of objects themselves. Relationships between objects are frequently mentioned in queries of tasks like semantic image retrieval, image captioning, visual question answering and natural language object detection. Hence, it is essential to model context between objects for solving these tasks. In the first part of this thesis, we present a technique for detecting an object mentioned in a natural language query. Specifically, we work with referring expressions which are sentences that identify a particular object instance in an image. In many referring expressions, an object is described in relation to another object using prepositions, comparative adjectives, action verbs etc. Our proposed technique can identify both the referred object and the context object mentioned in such expressions. Context is also useful for incrementally understanding scenes and videos. In the second part of this thesis, we propose techniques for searching for objects in an image and events in a video. Our proposed incremental algorithms use the context from previously explored regions to prioritize the regions to explore next. The advantage of incremental understanding is restricting the amount of computation time and/or resources spent for various detection tasks. Our first proposed technique shows how to learn context in indoor scenes in an implicit manner and use it for searching for objects. The second technique shows how explicitly written context rules of one-on-one basketball can be used to sequentially detect events in a game.

EACS: Effective Avoidance Combination Strategy

International audience

Challenges in imaging and predictive modeling of rhizosphere processes

Background Plant-soil interaction is central to human food production and ecosystem function. Thus, it is essential to not only understand, but also to develop predictive mathematical models which can be used to assess how climate and soil management practices will affect these interactions. Scope In this paper we review the current developments in structural and chemical imaging of rhizosphere processes within the context of multiscale mathematical image based modeling. We outline areas that need more research and areas which would benefit from more detailed understanding. Conclusions We conclude that the combination of structural and chemical imaging with modeling is an incredibly powerful tool which is fundamental for understanding how plant roots interact with soil. We emphasize the need for more researchers to be attracted to this area that is so fertile for future discoveries. Finally, model building must go hand in hand with experiments. In particular, there is a real need to integrate rhizosphere structural and chemical imaging with modeling for better understanding of the rhizosphere processes leading to models which explicitly account for pore scale processes.

Effects of fibrosis morphology on reentrant ventricular tachycardia inducibility and simulation fidelity in patient-derived models

Myocardial fibrosis detected via delayed-enhanced magnetic resonance imaging (MRI) has been shown to be a strong indicator for ventricular tachycardia (VT) inducibility. However, little is known regarding how inducibility is affected by the details of the fibrosis extent, morphology, and border zone configuration. The objective of this article is to systematically study the arrhythmogenic effects of fibrosis geometry and extent, specifically on VT inducibility and maintenance. We present a set of methods for constructing patient-specific computational models of human ventricles using in vivo MRI data for patients suffering from hypertension, hypercholesterolemia, and chronic myocardial infarction. Additional synthesized models with morphologically varied extents of fibrosis and gray zone (GZ) distribution were derived to study the alterations in the arrhythmia induction and reentry patterns. Detailed electrophysiological simulations demonstrated that (1) VT morphology was highly dependent on the extent of fibrosis, which acts as a structural substrate, (2) reentry tended to be anchored to the fibrosis edges and showed transmural conduction of activations through narrow channels formed within fibrosis, and (3) increasing the extent of GZ within fibrosis tended to destabilize the structural reentry sites and aggravate the VT as compared to fibrotic regions of the same size and shape but with lower or no GZ. The approach and findings represent a significant step toward patient-specific cardiac modeling as a reliable tool for VT prediction and management of the patient. Sensitivities to approximation nuances in the modeling of structural pathology by image-based reconstruction techniques are also implicated.

Scene Mapping and Understanding by Robotic Vision

The first mechanical Automaton concept was found in a Chinese text written in the 3rd century BC, while Computer Vision was born in the late 1960s. Therefore, visual perception applied to machines (i.e. the Machine Vision) is a young and exciting alliance. When robots came in, the new field of Robotic Vision was born, and these terms began to be erroneously interchanged. In short, we can say that Machine Vision is an engineering domain, which concern the industrial use of Vision. The Robotic Vision, instead, is a research field that tries to incorporate robotics aspects in computer vision algorithms. Visual Servoing, for example, is one of the problems that cannot be solved by computer vision only. Accordingly, a large part of this work deals with boosting popular Computer Vision techniques by exploiting robotics: e.g. the use of kinematics to localize a vision sensor, mounted as the robot end-effector. The remainder of this work is dedicated to the counterparty, i.e. the use of computer vision to solve real robotic problems like grasping objects or navigate avoiding obstacles. Will be presented a brief survey about mapping data structures most widely used in robotics along with SkiMap, a novel sparse data structure created both for robotic mapping and as a general purpose 3D spatial index. Thus, several approaches to implement Object Detection and Manipulation, by exploiting the aforementioned mapping strategies, will be proposed, along with a completely new Machine Teaching facility in order to simply the training procedure of modern Deep Learning networks.

Big Code Applications and Approaches

The availability of a huge amount of source code from code archives and open-source projects opens up the possibility to merge machine learning, programming languages, and software engineering research fields. This area is often referred to as Big Code where programming languages are treated instead of natural languages while different features and patterns of code can be exploited to perform many useful tasks and build supportive tools. Among all the possible applications which can be developed within the area of Big Code, the work presented in this research thesis mainly focuses on two particular tasks: the Programming Language Identification (PLI) and the Software Defect Prediction (SDP) for source codes. Programming language identification is commonly needed in program comprehension and it is usually performed directly by developers. However, when it comes at big scales, such as in widely used archives (GitHub, Software Heritage), automation of this task is desirable. To accomplish this aim, the problem is analyzed from different points of view (text and image-based learning approaches) and different models are created paying particular attention to their scalability. Software defect prediction is a fundamental step in software development for improving quality and assuring the reliability of software products. In the past, defects were searched by manual inspection or using automatic static and dynamic analyzers. Now, the automation of this task can be tackled using learning approaches that can speed up and improve related procedures. Here, two models have been built and analyzed to detect some of the commonest bugs and errors at different code granularity levels (file and method levels). Exploited data and models’ architectures are analyzed and described in detail. Quantitative and qualitative results are reported for both PLI and SDP tasks while differences and similarities concerning other related works are discussed.