Logo

Publikacije (31)

Nazad

Visually impaired person might find it very difficult to locate an object that has been even slightly misplaced from its usual position. Unfortunately this is very common situation in a shared environment where multiple individuals can affect object’s position and where visually impaired person cannot rely on object’s position remaining unchanged since the last interaction with the object. In order to independently localize the object of its interest visually impaired person must rely on assistive technology. It is yet very unlikely that any single wearable assistive device will encompass the whole range of object localization scenarios and be universally adoptable to a broad range of environments. In this paper we propose indoors test setup for visually impaired persons by creating RFID based assistive environment – Visual Impairment Friendly RFID Room. The test setup can be used to evaluate RFID object localization and its use by visually impaired persons.

Visual impairment severely constraints the ability to independently conduct many everyday tasks that we usually do not consider challenging. Although some types of visual impairment can be treated efficiently there is still a considerable number of visually impaired persons, especially among aging population, which depend on help of others or assistive technology to sustain their life quality. Visually impaired person cannot perceive the full extent of surrounding information due to the lack of visual details. However great progress can be achieved if surrounding information can be somehow visually transformed to the subset of visual information that visually impaired person can perceive. To certain extent every impairment has individual characteristics, as different individuals may better respond to different subsets of visual information. Thus any assistive solution aiming to visually transform surrounding information to accommodate broad range of impairment conditions must be personalized in order to be effective. Virtual reality enables individuals to experience imaginary surroundings by tricking their visual senses and such virtual surroundings can be personalized to any extent desired. We use virtual reality, image processing, and RFID to create a test setup able to simulate visual impairment and visually transformed surroundings suitable for visual Impairment studies. The test setup enables gathering information and conducting studies on both healthy and visually impaired individuals in a controlled environment enabling reliable assistive technology development and testing.

E. Buza, Amila Akagić, Ingmar Bešić

Automated detection of asphalt pavement distresses is a very popular computer vision and image processing problem. In recent years, automated detection is an essential part of every pavement management system, since it allows very fast detection of distresses on the road. This is important because timely detection can prevent many road accidents, and hence it has potential to save lives. In this paper, we presented a new unsupervised image processing method for segmentation of the most common road distresses-pavement cracks. The method first performs slicing of an image into M×N sub-images, and then removes sub-images without cracks based on empirically defined threshold. Analysis is then carried out only on a small number of sub-images, which significantly reduces computation time. Then, a series of images processing tasks are performed to select only pixels with pavement cracks. The method is suitable as a pre-processing step in a number of computer vision tasks, and can provide rough estimation of damaged area in an image.

Ingmar Bešić, E. Buza, Razija Turcinhodzic

Team performance depends on both individual and collaborative skills. This dependence creates increasing education and training demand while striving to improve teams’ efficiency. Consequentially, training and education systems emerge with new capabilities that are changing the learning landscape. With increasingly disperse and mobile teams it can be very inefficient and costly to provide training and education in a centralized instructor-led classes’ manner. Remote solutions are able to reach far more potential users at any moment, and tend to be satisfactory and possibly preferred in many different training and education areas. Computer Aided Design (CAD) requires high quality graphics for positive impact and high satisfaction. Software tools used for hands-on CAD training exercises depend on dedicated Graphics Processing Unit (GPU) to deal with complex graphics processing needed to visualize virtual models in real-time. It is challenging for the remote training system to offer this high level of experience to remote users. In this paper, synergy of conventional CAD laboratory workstations into cells is proposed in order to create a cost-effective team training remote system. The system utilizes existing capabilities of dedicated GPUs and custom software modules to capture video, perform hardware HEVC encoding, and stream the video at low bitrates and sub-second latency to remote team members.

Color vision deficiency is a surprisingly frequent vision impairment, but not considered to be a mayor eye disease due to being inherited condition and not progressive condition. However it poses serious restrictions on a visually impaired person because vision deficiency tests are commonly used to disqualify individuals affected by color vision deficiency from certain occupations. Color vision deficiency cannot be cured, thus it is important to develop suitable assistive technology to overcome the restrictions it poses. Virtual reality can project custom and separate images to both eyes in a real-time and thus enabling a new class of assistive technology that can deliver visual information in a highly customized manner. Virtual reality based assistive technology is promising for age-related macular degeneration, diabetic retinopathy and particularly for color vision deficiency. Virtual reality prototype is created based on a video see-through setup using commercial virtual reality headset and stereo camera. The prototype uses custom image processing to transform visual information from the camera to color vision deficiency friendly form. Time-domain color mapping real-time image processing is proposed to improve scores on standard color vision deficiency tests - Ishihara tests. Experiment is conducted to evaluate a protanope time-domain color mapping with sinusoidal envelope.

Online courses have potential to reach broader audience when compared to traditional learning methods, as they can be made instantly available to groups of students dispersed over wide range of locations and time zones. Some learning topics are easily adopted to this online environment, while others may be challenging to implement as online courses due to their specific requirements and associated cost. If mastering certain topic requires hands-on experience then online course must be made interactive to match the experience of a traditional instructor-led class. Otherwise, online course will be perceived as unsatisfactory and lacking positive impact. Computer Aided Design (CAD) training requires both high performance Graphics Processing Unit (GPU) and hands-on experience with specific CAD software. Both these requirements are difficult to meet on a remote student computer without additional cost. In this paper, grouping CAD workstations into cells is proposed in order to perform required graphics processing using existing hardware and software resources available in an average CAD laboratory. Proposed cell framework uses video capture and GPU hardware encoding to stream the content to the remote students while capturing their interactive feedback for the near real-time hands-on experience. The framework requires single capture card per cell and utilizes 20% of the resources for the cell overhead processing. Remaining 80% of workstations are fully available to the online students and instructors.

E-Learning solutions for Computer Aided Design (CAD) require high quality graphics for positive impact and high satisfaction. Software tools used for hands-on CAD laboratory exercises depend on dedicated Graphics Processing Unit (GPU) to deal with complex graphics processing needed to visualize virtual models in real-time. It is challenging to offer the same level of experience to remote users as they usually cannot afford laboratory-level hardware with dedicated GPU nor such hardware can be provided to them as part of their e-Learning experience. In this paper, grouping of average CAD laboratory workstations in pairs is proposed in order to create remote stations capable of performing required CAD graphics processing for remote users while streaming it over the network for a near real-time experience. Remote station captures video and utilizes hardware HEVC encoding, as common capability of high-end dedicated GPUs, to perform low bitrate video streaming with sub-second latency. The remote station concept enables cost-effective extension of average conventional CAD laboratory to an e-Learning remote laboratory for up to 50% additional remote users in respect to the total number of the laboratory workstations.

Development of a spatial decision support system requires integration of various spatial data sets coming from different information systems of possibly more than one organization. The spatial decision support system development for spatial planning and environmental protection is discussed and spatial data integration is described. The heterogeneity of information systems from which spatial data come is reflected through their purpose. Utility and real estate cadaster systems imply services to citizens as part of spatial data infrastructure in the Federation of Bosnia and Herzegovina. The spatial planning information system is intended for registration of environmental changes and spatial decision making. Utility cadaster information system development is described along with important aspects on how to satisfy both functional cadastral services and spatial planning experts needs to analyze information related to land use and network supply systems. Particular attention is given to spatial data transformation for utility cadaster database development in accordance with a prescribed data model.

Estimator algorithms rely on assumed laser stripe image profile to determine its peek with sub-pixel accuracy. They depend on light intensity readings around the peak and are susceptible to noise and saturation. Noise and stripe intensity models are commonly used to synthesize and feed test data to estimator algorithms in order to evaluate their accuracy and robustness. For real-time 3D scanning applications estimator algorithms are expected to prefer less computationally demanding estimation techniques. Simple and accurate models of empirical noise and laser stripe profile could be used to improve testing and algorithms accuracy. Modular test setup for 3D scanning is utilized to project a laser stripe on the target with patterned surface. Laser stripe image is captured and processed to extract noise and surface pattern interference. Laser power modulation is used to generate series of captures with various stripe intensities. Captures are partitioned, analyzed and presented according to target surface properties and color channels. Image noise interfering with sub-pixel peak detection is analyzed and noise model based on empirical data is proposed. Empirical laser stripe images are analyzed and novel simple laser stripe intensity profile model conforming to empirical data is proposed.

Laser stripe sub-pixel peak estimator algorithms depend on the light intensity readings around the peak and are susceptible to saturation and noise. Real-time 3D scanning prefers fast estimator algorithms based on simple and yet accurate stripe models. Modular test setup for 3D scanning is utilized to project a laser stripe on the target with patterned surface. Laser power modulation is used to generate series of captures with various stripe intensities. Stripe images are analyzed to find preferred modulation range suitable for estimator algorithms on both black and white surfaces. Empirical noise and laser stripe profile data are used to verify estimator algorithms assumptions. A novel algorithm for sub-pixel peak detection applicable to saturated laser line images is proposed.

Laser triangulation 3D scanning is performed on a setup that closely matches intended target size, desired spatial resolution, intended environment and target surface types. Modular test setup is proposed to enable laser triangulation 3D scanning using exchangeable sets of components supporting wider range of possible scanning options. Software framework is proposed to programmatically control the setup in a unified manner independently on underlying components used. Automated test runs are carried out, visualized, saved and repeated under desired conditions using command interpreter and batches of commands. Proposed software framework divides image processing into sequential modular blocks convenient for pipeline execution. Light pattern extraction block enables camera focus and laser source modulation feedback to be incorporated into image processing algorithms under test.

The purpose of this study is to present novel GAANFIS expert system prototype for tar detection in cigarettes during manufacturing process. The proposed system combines capabilities of Adaptive Neuro-Fuzzy Inference System (ANFIS) and Genetic Algorithm (GA).The data recorded for different type of cigarettes are collected by special control quality equipment in real conditions inside cigarette factory. GA-ANFIS system performs optimization in two steps. In the first step it generates six different ANFIS structures, and after that, we have second level of GA optimization using given ANFIS structures resulting in optimal fuzzy model structure. Modeling and validation of the GA-ANFIS system approach is performed in MATLAB environment using validation data set that were not used in the process of training. Our earlier research results based on two different approaches (ANFIS and High-performance liquid chromatography (HPLC)) are also introduced. Performances of these three approaches are compared and novel expert system prototype shows better result related to training, testing and validation errors. It also more precisely shows that low yield tar cigarettes contain similar levels of nicotine opposite to high yield tar cigarettes while benzene, toluene, and xylene (BTX) levels rise along with increasing tar yields.

L. Fazlic, A. Avdagic, Ingmar Bešić

The aim of this research is to develop a novel GA-ANFIS expert system prototype for classifying heart disease degree of a patient by using heart diseases attributes (features) and diagnoses taken in the real conditions. Thirteen attributes have been used as inputs to classifiers being based on Adaptive Neuro-Fuzzy Inference Systems (ANFIS) for the first level of fuzzy model optimization. They are used as inputs in Genetic Algorithm (GA) for the second level of fuzzy model optimization within GA-ANFIS system. GA-ANFIS system performs optimization in two steps. Modelling and validating of the novel GA-ANFIS system approach is performed in MATLAB environment. We compared GA-ANFIS and ANFIS results. The proposed GA-ANFIS model with the predicted value technique is more efficient when diagnosis of heart disease is concerned, as well the earlier method we got by ANFIS model.

Many different applications like face/people detection, image content interpretation, de-identification for privacy protection in multimedia content, etc. requires skin detection as a pre-processing step. There is no a perfect solution for skin detection, since this process is a compromise on speed, simplicity and precision (detection quality). There are many different techniques for skin detection modeling ranging from simple models based on one or several thresholds to advanced models based on neural network, Bayesian classifier, maximum entropy, k-means clustering, etc. This paper proposes a simple model, based on ratios of red, green and blue components of the RGB color model. It describes how to make a compromise in a skin detection modeling by using three levels of rules. Data analysis that supports conclusions is performed on the dataset from Universidad de Chile (UChile, dbskin2 - complete set) that contains 103 images and their annotations.

This paper presents two approaches in isolation of vibrations of driver's seat . The first approach shows the ability of Fuzzy Logic Controller (FLC) to adjust the stiffness of the air spring, which is implemented between cabin floor and the seat together with damper in order to isolate the vertical vibrations. The second approach is based on Artificial Neural Network Controller (ANNC) with purpose of improvement vibration isolation producing appropriate voltage for valve flow diameter control of semi-active damper. The quality of isolation is measured using standardized technique. The results of simulation in Matlab/Simulink, as well as the results of implemented controllers on a real experimental model are presented.

Nema pronađenih rezultata, molimo da izmjenite uslove pretrage i pokušate ponovo!

Pretplatite se na novosti o BH Akademskom Imeniku

Ova stranica koristi kolačiće da bi vam pružila najbolje iskustvo

Saznaj više