Intelligent Interactive Multimedia Systems and Services

The paper deals with target selection techniques for wearable augmented reality systems. In particular, we focus on the three techniques most commonly used in distant freehand pointing and clicking on large displays: wait to click, air tap and thumb trigger. The paper details the design of the techniques for a touchless augmented reality interface and provides the results of a preliminary usability evaluation carried out in out-of-lab settings.

Internet of Things (IoT) computing applied to the Cultural Heritage domain is an emerging discipline which consists of the application of intelligent sensors and technologies within cultural sites; it is strongly related to the development of systems able to be pervasive and ubiquitous with the definitive goal of rethinking such spaces. IoT paradigm can constitute a powerful tool to enhance people fruition and enjoyment of such spaces; thanks to ICT technologies, a cultural object can be effectively “dressed” of its context and juxtaposed into it. In this paper, an intelligent IoT system, designed with the aim of improving user experience and knowledge diffusion within a cultural space, is presented. The paper describes the hardware/software system components, and presents a case study of a sculptures exhibition named

the Beauty or the Truth

(

http://www.ilbellooilvero.it

) in Naples where the system was deployed. Furthermore, the paper provides the results of an users behaviour analysis which revealed up a significant increase in user satisfaction and cultural knowledge diffusion.

We present a multimedia summarizer system for retrieving relevant information from some web repositories based on the extraction of semantic descriptors of documents. In particular, semantics attached to each document textual sentences is expressed as a set of assertions in the

$$\langle subject,verb,object \rangle $$

shape as in the RDF data model. While, images’ semantics is captured using a set of

keywords

derived from high level information such as the related title, description and tags. We leverage an unsupervised clustering algorithm exploiting the notion of semantic similarity and use the centroids of clusters to determine the most significant summary sentences. At the same time, several images are attached to each cluster on the base of keywords’ term frequency. Finally, several experiments are presented and discussed.

Three-dimensional data acquisition systems are becoming progressively more affordable, especially those that rely on photographic cameras and motion sensing input devices such as the Microsoft Kinect and PlayStation Eye. At the same time, mobile devices capable of rendering complex 3D graphics and with always-on broadband connectivity are becoming increasingly wide-spread. As a result, there is a potential opportunity for the development of novel mobile-oriented streaming mechanisms that can support the visualisation of large and rapid changing spatial-temporal datasets. In this work, we introduce a framework for the visualisation of spatial-temporal point cloud models on mobile devices. One advantage of our framework is that it eliminates the need to pre-downloading spatial-temporal models prior to their visualisation, hence avoiding the need of large storage requirements. Additionally, it streams spatial-temporal geometry using progressive levels of detail that are optimised to the mobile’s rendering capabilities and network bandwidth. The representation of different levels of detail is not bounded to a particular geometry reduction algorithm and the streaming process is totally transparent to the user, who perceives remote 3D models as local ones. To evaluate the relevance and impact of this study, a use case is also presented.

This paper proposes a method to verify the facial shape model. The method can be used to improve or use of the shape model. We verify the shape model by analyzing the texture and shapes of landmarks that constitute the outline of the shape. It generates the facial face image of the face that is tracked for the invariant verification method to poses of the face. Next, it reduces the effect of environment by configuring a patch surrounding each landmark. The texture of the skin area to these patches is analyzed by the gray-value variance. Then, the result of analysis is corrected by the relationship between the landmarks of the shape. As a result, we identified the fitting result as true or false to the Multi-PIE database and obtained the accuracy of 83.32 %.

Data analysis and monitoring is currently carried out within enterprises using Business Intelligence tools that are subject to major limitations (as outlined in the state of the art analysis that we perform). Effective visualization support is a very much needed feature in Big Data applications. In this paper we examine the visualisation requirements of a real world banking application, and identify generic visualisation tasks that are essential for doing effective analysis of a complex process that produces amazingly large amounts of data. The requirements for the visualization support that we propose are modelled using an application wireframe that acts a story-board. The effectiveness of the visualization facilities that we propose is demonstrated through their application to the Big Data banking use-case.

This study evaluates, the effectiveness of pictogram meanings for order picking processes with disabled people and people with altered performance. Pictograms were created for guidance and feedback of process steps in order picking. The comprehensibility and colors of the pictograms were evaluated through a survey with 45 normal performance people and 26 disabled people. The interviewees responded to a standardized questionnaire with at least three different types of pictograms for each process step.

On line social networks (e.g., Facebook, Twitter) allow users to tag their posts with geographical coordinates collected through the GPS interface of smart phones. The time- and geo-coordinates associated with a sequence of tweets manifest the spatial-temporal movements of people in real life. This paper aims to analyze such movements to discover people and community behavior. To this end, we defined and implemented a novel methodology to mine popular travel routes from geo-tagged posts. Our approach infers interesting locations and frequent travel sequences among these locations in a given geo-spatial region, as shown from the detailed analysis of the collected geo-tagged data.

In the last decade, the World Wide Web has been evolving as a data infrastructure, where a wide variety of resources is increasingly being made available as Web services. This trend is pushing the researchers to investigate approaches like composition platforms, aimed at empowering end users to access, compose and use these services. Despite the wide availability of data sources, due to the specific and diverse end users’ information needs often no data source can satisfy these needs. This limits the adoption of composition platforms in real contexts and everyday use. In order to overcome this limitation, this paper presents a polymorphic data source that exploits the wide availability of information structured in the Linked Open Data cloud. To build this data source, a semi-automatic annotation algorithm is presented that creates semantic annotations for services available in a composition platform. An implementation of this approach in a mashup platform is described.

In this paper, we show that rational decision-making methods such as Benjamin Franklin’s can be successfully implemented as a text based natural language dialog system. More specifically, we developed a prototype,

vpino

, and conducted a user study. Vpino acts maieutically: the questions raised by vpino encourage the user to reflect about potential options and arguments and help her structure her thoughts. To maintain a real dialog,

vpino

unobtrusively attempts to keep control of the conversation at all times. Serious, motivated users evaluated acceptance and usefulness of

vpino

quite positively. Users that are more open to computer based decision support held better and more fruitful dialogs than those with a sceptical attitude. This quantitative result conforms well to our qualitative observation that

vpino

shows good human like behaviour whenever the user is serious and motivated. We also found that users with a more hypervigilant approach to decisions particularly benefit from

vpino

.

E-Health organisations have seen in these years a rapid growth in the complexity and criticality of the processes they manage. This paper defines an approach for modelling clinical workflows based on Model-Driven principles; in particular the modelling activity is supported by the Dynamic State Machine (DSTM) formalism, that is a well-formed graphical language able to represent state based systems. The main advantage of using such a language resides in obtaining formal models of clinical workflows (whose semantics is strong and precise), with an high level of usability. While the focus of the paper is clearly on modelling, the application of Model-Driven principles allows a tight integration between the control flow of the clinical processes and the information that can be extracted from informal documentation. The approach is shown by applying it to the case study of a real world treatment process of bipolar and mood disorders.

Alzheimer’s disease (AD) is the most common type of dementia; accounts report the impact of AD in a range of 60–80 % with respect all the dementia related pathologies. Patients with Alzheimer’s could show early symptoms such as sleep disturbances, well-formed visual hallucinations, and muscle rigidity or other AD movements disorders. In this work we focus on disturbances related to sleep disorders, for this reason, we propose a monitoring system for the recognition of these kind of disorders. In detail, we present: (1) a novel model for the detection of a particular type of sleep disorders, the Periodical Legs Movements (PLM) and (2) a prototype of monitoring system able to recognize the PLM events. Preliminary experiment have been made by monitoring a patient with a early stage of the sleep disorder.

This paper presents a modified Paley method for calculation of Mersenne matrices at order values equal to odd prime numbers. Some examples of Mersenne matrix sorting, allowing for calculation of the complete set of functions, are also considered. A comparison of Walsh and Mersenne-Walsh systems of functions in terms of their properties and fields of application is provided. The efficiency of this topic for use in the development of band-pass filters is indicated.

A technique of video sequence processing through the energy parameters analysis of video signal is discussed in this paper. The purpose of the analysis here is to determine the object movements in a real-life scene. The new technique is theoretically justified. It is based on the dependence of energy signal at the initial phase. The signal is represented as a spatial cosine segment. The differences of the energy characteristics are introduced to improve the efficiency of video sequences processing. An example of a video sequence processing of the real scene is considered. It is shown that if pixels orderly move inside the object, then the borders and the velocity of the moving object are well determined by using the characteristics difference. However, the pixels groups with the movement differ from the object movement can exist within the object boundaries. In this case, the pixels must be regularly zeroed out, starting from the vision area edges.

The article discusses a significant expansion of the basis of two-level quasi-orthogonal Hadamard-Mersenne matrices, used for compression and masking of images and video stream frames. The article presents the dependences of the levels of matrix elements on their order, and the algorithm for calculation through chains that include Hadamard-Euler matrices. It is demonstrated that the sizes of Hadamard and Hadamard-Mersenne matrix families are the same. The procedure of masking and compression of images using quasi-orthogonal matrices, the orders of which correspond to the sequence of Mersenne numbers, is analyzed.

Performance analysis of several state-of-the-art prediction approaches is performed for lossless image compression. To provide this analysis special models of edges are presented: bound-oriented and gradient-oriented approaches. Several heuristic assumptions are proposed for considered intra- and inter-component predictors using determined edge models. Numerical evaluation using image test sets with various statistical features confirms obtained heuristic assumptions.

In this paper the problem of correlation noise modeling (CNM) for Transform Domain Wyner-Ziv (TDWZ) video coding is considered. The CNM algorithm from DISCOVER codec is analyzed. An extended set of assumptions about correlation noise in transform domain is considered and a new CNM algorithm is proposed. The proposed algorithm is a generalization of the reference one for the case of non-stationary noise. The evaluation results show that the proposed method demonstrates higher performance than the state-of-the-art approach on test video sequences with complex motion.

During airborne laser scanning, different types of information are available including Light Detection And Ranging (LiDAR) data as a cloud of 3D points, aerial digital photography data (hyperspectral or color visual images), and additional information about parameters of shooting. Difficulties of large image stitching due to parallax effects lead to distortions between ground truth 3D LiDAR coordinates and 2D visual coordinates of the same point. Our contribution is to develop a method for fusion of LiDAR and visual information for accurate segmentation of individual tree crowns in order to receive biomass measurements. The shearlet theory was used to improve boundaries and texture of airborne images. Also in this paper, a higher-order active contour model is applied for area evaluation of tree crowns in a plane. The received area measurements are promising and coincide with expert estimations providing accuracy 92–96 %. The modeling results are good for non-Lambert space of forest.

In this paper, a technical solution of digital gray-scale watermarking for law enforcement and copyright protection for digital media is developed. The biometrically generated keys provide higher level of security, when a key is created using biometrical information, in our case—the fingerprint images. First, feature points and, second, fast binary descriptors are applied for secret keys creation. Also a map of feature points is used to randomize the embedded watermark. A blind scheme of gray-scale watermark transform is implemented, when the secret keys ought to be transferred through a secret channel without necessity to transfer whole host image. Two types of transforms such as Discrete Cosine Transform (DCT) and Discrete Wavelet Transform (DWT) are investigated for embedding of watermark in a host image. Numerical estimations of attacked image restoration show a superiority of the DWT for geometric attacks including shifts and rotations.

In the article a new non-intrusive speech quality assessment algorithm for packet switched wireless systems is proposed. The new method is based on PESQ and E-model algorithms and provides quite good accuracy. To estimate performance of the algorithm a model of the 3GPP LTE system is considered. Simulations show that correlation of the results received from the proposed algorithm and PESQ for anchor modes of AMR-NB and AMR-WB speech codecs is 97 and 94 %, respectively.

In the paper the usage of low-density parity-check (LDPC) codes to protect storage systems from failures is considered. These codes are the instance of locally recoverable (LRC) codes which obtain much attention during last years regarding storage systems. The system model of distributed storage system is described, with specific types of failures. The coding schemes based on Reed-Solomon (RS) and LDPC codes are formulated for this model taking into account the specific failures types. These coding schemes are compared using several examples of model parameters. The redundancy and locality provided by different coding schemes are estimated.

This article considers transport coding which is a method for data transmission in a packet switching network. It uses error correcting codes at the transport layer of data network and can help to decrease the mean message delay. Evaluation of the mean message delay is important for estimation of transport coding efficiency. The existing analysis of the mean message delay uses assumption about exponential distribution of packet delay. The network model proposed by L. Kleinrock is considered in this paper. The distribution of the packet delay in Kleinrock network is not proven to be exponential. This work offers calculation of the mean message delay for Kleinrock networks without usage of the assumption about exponential distribution of the packet delay. The accuracy is checked using simulation of the Kleinrock network model.

It was suggested [

1

] that having natural frequency of the front approximately 80 % of that of the rear suspension in a vehicle will result in a flat ride for the passengers. Flat Ride in this case means that the pitch motion of the vehicle, generated by riding over a bump for instance will fade in to the bounce motion of the vehicle much faster. Bounce motion of the vehicle in mush easier to tolerate and feels more comfortable for the passengers. In a previous study the authors, analytically proved that this situation is not practical. In other words, for any vehicle there will only be one certain velocity, depending on the geometry and suspension system specifications which the flat ride will happen at. The search continued to find a practical method for enjoying the flat ride in vehicles. Solving the equation of motion of the vehicle for different spring rates and road configuration the authors came up with design chart for smart suspension systems. Using the advantages of the analytical approach to the flat ride problem, the chart was established to be used for vehicles with smart active suspension systems. In this paper the mathematical methods used and the resulted criteria for designing a flat ride suspension system which will perform in different speeds is presented.

During the project for generating a mathematical algorithm for autonomous vehicles, a sample road which included different turns and scenarios was required. Studying different types of roads and their line equations Euler spirals, also known as Clothoids were found to be the best solution for designing new roads suitable for autonomous vehicles. During 19th century Arthur Talbot derived the equation of Clothoids to be used as an easement curve for the purpose of avoiding shock and disagreeable lurch of trains, due to instant change of direction. The Euler Spiral is a curve whose degree-of-curve increases directly with the distance along the curve from the start point of the spiral. This provides a linear change in the steering angle required by the driver to go through the turn. In other words for a car traveling on a Clothoid transition road curve there is no need for sudden changes in the steering angle of the wheels. The angle required starts from zero and increases to a maximum value and goes back to zero linearly. This provides a very comfortable ride for the passengers of the vehicle. The use of these curves for road design have been investigated, and a design chart have been proposed to be used for finding the best suitable transition curve for different applications.

Autonomous systems perform decision making without human intervention. They collect the data from the environment, process it to build the awareness and perform the actions. Consequently, the adaptivity functions can be seen as the core parts of these systems. In this paper we introduce a formal framework for modelling and analysis of autonomous systems and their compositions, especially focusing on the adaptivity modelling aspects and reasoning about adaptive behaviour.

Design solution of a novel mobile robot navigation system, presented here, is used to control robot’s locomotion across slippery surfaces. Usually, motion control strategies, are based on assumption of sufficient traction between tyres and the road. Motion across slippery surfaces can endanger the robot and its surroundings. Our solution combines Light Detection and Ranging (LIDAR) measurements with odometry data. It performs well on any surface, regardless of sensing, localization and navigation errors, within an indoor environment, in real-time. An accelerated feature detection method is used to improve LIDAR localization update rate and improve localization accuracy. Experiments conducted validate proposed approach.

The benefits of stationary aerial platforms for continuous observation of a fixed area are evident in cases such as natural disasters, environmental monitoring, intelligence gathering, etc. Lighter-Than-Air (LTA) technology has advanced significantly, for example high-pressure balloons that would make such a platform a potential solution for long, uninterrupted and persistent observation. An adaptive nonlinear control system was designed for LTA station-keeping, with an adaption mechanism to accommodate uncertainties in system parameters. The controller was subsequently implemented on a micro controller and tested in a flight experiment using a helium balloon. The tests were conducted in a 4.5 m by 4.5 m enclosure using infrared sensors to determine position by measuring distance from the walls.

A user of existing Content-Based Image Retrieval (CBIR) systems suffers from single query key selection, in case that he/she might begin retrieval process toward an ambiguous goal. In the other case, he/she cannot judge whether one image is more adequate than the other exactly for retrieval. To overcome the difficulty, this paper proposes Multi-Key CBIR (MK-CBIR) which differs from Single-Key CBIR (SK-CBIR) in retrieving similar images from image databases with a multi-key query, not a single-key one. To implement MK-CBIR,

skyline

query based MK-CBIR (

$$\Phi _{sky}$$

Φ

s

k

y

) and aggregate

k

-Nearest Neighbor query based MK-CBIR (

$$\Omega _{sum}$$

Ω

s

u

m

,

$$\Omega _{max}$$

Ω

m

a

x

, and

$$\Omega _{min}$$

Ω

m

i

n

) are presented. Experimental results on retrieval performance of MK-CBIR are as follows:

Recall

of

$$\Omega _{max}$$

Ω

m

a

x

is over 0.3 higher than that of

k

-NN search and the highest among those of MK-CBIR. It is comparatively stable against noisy keys.

Recall

of

$$\Omega _{min}$$

Ω

m

i

n

is roughly the same as that of

k

-NN search and the lowest among those of MK-CBIR.

Recall

of

$$\Omega _{sum}$$

Ω

s

u

m

and

recall

of

$$\Phi _{sky}$$

Φ

s

k

y

are roughly the same and between those of

$$\Omega _{min}$$

Ω

m

i

n

and

$$\Omega _{max}$$

Ω

m

a

x

.

It is concluded that MK-CBIR is better than SK-CBIR in retrieval performance.

We propose a simple medium access scheme for wireless ad hoc network (WAHN). The proposed method is based on the spread spectrum (SS) technique to take advantage of the robustness against fading and differentiation of signals. However, conventionally, SS needed a mechanism to allocate a unique spreading code to each node. Since WHAN does not include base stations, this problem must be solved. In this paper, we solve this problem by let each node share a unique code and use it after expanding the chip duration. Although this proposed scheme decreases the data rate if the node expands widely the chip duration, it is better than CSMA. The performance is verified through computer simulations.

This paper develops a new Arduino-compatible board for sensor networks and proposes a simple tree-based data collection scheme for low-power microcontrollers. The developed board has a special circuit for a stabilized power source for external sensors and wireless modules, and implements a charge controller from a solar cell to Li-Po battery for autonomous long-lived operation. The implemented tree-based data collection scheme consists of a routing function and a media access control function for low-power microcontrollers. We can apply it as a small size program which is executed in a small size of the execution area by reducing routing information, and it can construct a tree-based route from a sink node to each node on a hop by hop basis. The developed media access control function employs a special time-division frame structure to leverage a sleep operation with low-power consumption for long-lived operation, and to reduce packet collisions between adjacent nodes. The experimental evaluations show that the proposed data collection scheme can operate on the developed board with a low-power microcontroller even if the microcontroller has only 8 [KB] Static Random Access Memory (SRAM). Additionally, the implemented tree-based data collection scheme can construct an adequate route from nodes to a sink node, and can realize a reliable data collection.

Japanese road traffic act was revised in November, 2011. In the revised traffic act, the configuration of roundabout intersection and its traffic regulation are defined. The roundabout intersection is already widely used in Europe, but Japanese people are less familiar with it. According to the report of MLIT (Ministry of Land, Infrastructure, Transport and Tourism), there are only about 140 roundabout intersections in Japan, but most of them are narrow and small compared to Europe. Tentative examinations about the effect of the roundabout intersection were carried out in a few cities (e.g., Karuizawa-Cho and Iida-shi). The results of the examination indicate that the intersection environment can be improved by introducing the roundabout intersections. It seems that the roundabout intersection will become more popular in Japan. Therefore, we aim at analyzing the driving behaviors at roundabout intersections to model a virtual driver for traffic simulation. The traffic simulation based on the model will assist novice drivers to avoid accidents in roundabout intersections. In this paper, we construct a virtual driving course based on an existing roundabout intersection at Ichinomiya-shi, Aichi by using a driving simulator “UC-win/Road”. Moreover, we analyze the log data such as accelerator and brake pedals from some viewpoints.

In this paper, we discuss the effect of novel reading experience on composing technical papers and investigate how the novel reading experiences in which many students have been as their interesting activities should be applicable to the paper composition work. Our goals were to extract the common features between novels and technical papers in comparison with their different viewpoints, and then to find out the creative ability of paper writing from intelligent work in the novel reading process.

Our research goal is to develop evacuation guidance map systems for disaster situations such as major earthquakes. Our guidance system provides a safe-road map in disaster areas that is constructed based on ant colony systems. We describe a movement algorithm for evacuee agents utilizing an evacuation guidance system in a disaster simulator. In the simulators, evacuee agents move to safe shelters in accordance with information on the guidance system. We evaluated the effectiveness of the evacuation guidance systems using the movement algorithm. Experimental results show that the evacuee agents utilizing the guidance system can reach the shelters faster.