1. Background purification framework with extended morphological attribute profile for hyperspectral anomaly detectionJu Huang, Kang Liu, Mingliang Xu, Matjaž Perc, Xuelong Li, 2021, original scientific article Abstract: Hyperspectral anomaly detection has attracted extensive interests for its wide use in military and civilian fields, and three main categories of detection methods have been developed successively over past few decades, including statistical model-based, representation-based, and deep-learning-based methods. Most of these algorithms are essentially trying to construct proper background profiles, which describe the characteristics of background and then identify the pixels that do not conform to the profiles as anomalies. Apparently, the crucial issue is how to build an accurate background profile; however, the background profiles constructed by existing methods are not accurate enough. In this article, a novel and universal background purification framework with extended morphological attribute profiles is proposed. It explores the spatial characteristic of image and removes suspect anomaly pixels from the image to obtain a purified background. Moreover, three detectors with this framework covering different categories are also developed. The experiments implemented on four real hyperspectral images demonstrate that the background purification framework is effective, universal, and suitable. Furthermore, compared with other popular algorithms, the detectors with the framework perform well in terms of accuracy and efficiency. Keywords: detectors, anomaly detection, image reconstruction, hyperspectral imaging, training, optics, dictionaries, background purification, extended attribute profile, sparse representation, stacked autoencoder Published in DKUM: 19.08.2024; Views: 92; Downloads: 9 Full text (5,36 MB) This document has many files! More... |
2. Vine canopy reconstruction and assessment with terrestrial lidar and aerial imagingIgor Petrović, Matej Sečnik, Marko Hočevar, Peter Berk, 2022, original scientific article Abstract: For successful dosing of plant protection products, the characteristics of the vine canopies should be known, based on which the spray amount should be dosed. In the field experiment, we compared two optical experimental methods, terrestrial lidar and aerial photogrammetry, with manual defoliation of some selected vines. Like those of other authors, our results show that both terrestrial lidar and aerial photogrammetry were able to represent the canopy well with correlation coefficients around 0.9 between the measured variables and the number of leaves. We found that in the case of aerial photogrammetry, significantly more points were found in the point cloud, but this depended on the choice of the ground sampling distance. Our results show that in the case of aerial UAS photogrammetry, subdividing the vine canopy segments to 5 × 5 cm gives the best representation of the volume of vine canopies. Keywords: precision agriculture, remote sensing, 3D point clouds, vineyard, canopy reconstruction, terrestrial lidar, aerial photogrammetry, manual defoliation Published in DKUM: 15.07.2024; Views: 123; Downloads: 8 Link to full text This document has many files! More... |
3. |
4. |
5. Novel Half-Spaces Based 3D Building Reconstruction Using Airborne LiDAR DataMarko Bizjak, Domen Mongus, Borut Žalik, Niko Lukač, 2023, original scientific article Abstract: Automatic building reconstruction from laser-scanned data remains a challenging research topic due to buildings’ roof complexity and sparse data. A novel automatic building reconstruction methodology, based on half-spaces and a height jump analysis, is presented in this paper. The proposed methodology is performed in three stages. During the preprocessing stage, the classified input point cloud is clustered by position to obtain building point sets, which are then evaluated to obtain half-spaces and detect height jumps. Half-spaces represent the fundamental shape for generating building models, and their definition is obtained from the corresponding segment of points that describe an individual planar surface. The detection of height jumps is based on a DBSCAN search within a custom search space. During the second stage, the building point sets are divided into sub-buildings in such a way that their roofs do not contain height jumps. The concept of sub-buildings without height jumps is introduced to break down the complex building models with height jumps into smaller parts, where shaping with half-spaces can be applied accurately. Finally, the sub-buildings are reconstructed separately with the corresponding half-spaces and then joined back together to form a complete building model. In the experiments, the methodology’s performance was demonstrated on a large scale and validated on an ISPRS benchmark dataset, where an RMSE of 0.29 m was obtained in terms of the height difference. Keywords: LiDAR point cloud, building reconstruction, half-spaces, Boolean operations Published in DKUM: 01.12.2023; Views: 388; Downloads: 23 Full text (13,79 MB) This document has many files! More... |
6. Three-Phase load current reconstruction using singleshunt current measurement and modified vector modulation for three-level DC-AC converter : doctoral dissertationHaris Kovačević, 2022, doctoral dissertation Abstract: This doctoral thesis focuses on the development of Space Vector Modulation (SVM) techniques adjusted for single-shunt current measurement and applied to the three-level three-phase voltage source inverter. Based on the single-shunt current measurement, three-phase load currents are then reconstructed.
An overview of existing multilevel topologies and their main operating principles is provided within the thesis. To evaluate proposed SVM techniques adjusted for single-shunt current measurement, a three-level three-phase diode clamped voltage source inverter topology is selected. The main advantages of the selected topology are high efficiency, reduced semiconductor stress, the low capacitance of DC-link capacitors, etc. To reconstruct the three-phase load currents, a single-shunt is positioned inside the middle branch of the DC-link.
An overview of existing modulation techniques for two-level and multilevel inverters is provided within the thesis. To reconstruct three-phase load currents, three SVM methods are proposed within the thesis. Due to the lack of a current measurement window, while the voltage vector is positioned near sector or region boundary areas, additional modifications are required. The first SVM method shifts the SVM signals to ensure the minimum current measurement window needed for measurement with minimum additional vector injection. Such an approach ensures the high modulation index values greater than 0.8 with the smallest current ripple due to the SVM signal symmetry. The second method intentionally injects the colinear voltage vector into the existing vector combination to ensure the minimum time window required for current measurement. As a drawback, a smaller modulation index compared to the first method can be achieved. As an advantage, a very low modulation index of less than 0.2 can be achieved. The third method simplifies the second method by using a single SVM pattern. Such a method can be used only for very low modulation indexes and can be applied for two-level and three-level inverters.
The proposed SVM methods are implemented within the Digital Signal Processor (DSP). Experimental results confirm the proper operation of the proposed methods applied to the three-level diode clamped voltage source inverter. Keywords: single-shunt, three-level inverter, SVM modification, current reconstruction, vector injection Published in DKUM: 09.03.2023; Views: 396; Downloads: 55 Full text (12,21 MB) |
7. Capturing Conversational Gestures for Embodied Conversational Agents Using an Optimized Kaneda–Lucas–Tomasi Tracker and Denavit–Hartenberg-Based Kinematic ModelGrega Močnik, Zdravko Kačič, Riko Šafarič, Izidor Mlakar, 2022, original scientific article Abstract: In order to recreate viable and human-like conversational responses, the artificial entity, i.e., an embodied conversational agent, must express correlated speech (verbal) and gestures (non-verbal) responses in spoken social interaction. Most of the existing frameworks focus on intent planning and behavior planning. The realization, however, is left to a limited set of static 3D representations of conversational expressions. In addition to functional and semantic synchrony between verbal and non-verbal signals, the final believability of the displayed expression is sculpted by the physical realization of non-verbal expressions. A major challenge of most conversational systems capable of reproducing gestures is the diversity in expressiveness. In this paper, we propose a method for capturing gestures automatically from videos and transforming them into 3D representations stored as part of the conversational agent’s repository of motor skills. The main advantage of the proposed method is ensuring the naturalness of the embodied conversational agent’s gestures, which results in a higher quality of human-computer interaction. The method is based on a Kanade–Lucas–Tomasi tracker, a Savitzky–Golay filter, a Denavit–Hartenberg-based kinematic model and the EVA framework. Furthermore, we designed an objective method based on cosine similarity instead of a subjective evaluation of synthesized movement. The proposed method resulted in a 96% similarity. Keywords: conversational gestures, 3D gestures, motor skills, gesture reconstruction, kinematics, embodied conversational agents, Kanade–Lucas–Tomasi tracker, Denavit–Hartenberg Published in DKUM: 20.01.2023; Views: 622; Downloads: 75 Full text (5,04 MB) This document has many files! More... |
8. Revitalization and valorization of Onogošt fortress in Nikšić : master thesisĐina Prnjat, 2021, master's thesis Abstract: Fortress "Onogošt", known to the public simply as Bedem, is the prominent symbol of Nikšić and a valuable fragment of the cultural and historical heritage of Montenegro. In the past decade the fortress has been renovated several times, aiming to emphasize the value and importance of the old ramparts, as well as to evaluate the spectrum of possibilities of its adaptation in the context of contemporary public city space.
These interventions breathed new life into the dilapidated debris, implying its transformation from a warrior fortification into a reclaimed cultural and artistic center of the city. However, this idea never fully flourished, and the current condition of the city fortress is far from imagined- neglected, crumbled and devastated.
With the exception of regional music festival “Bedem fest”, a and a few minor manifestations that have only scratched the surface of its enormous potential occasionally in the past decade, there are almost no other active initiatives and this monumental space remains the silent neglected urban pocket in the heart of the city. Although still dominant and crucial in the visual perception of Nikšić, the fortress today represents essential ravaged resource of the city.
The purpose of this master thesis is to restore, conserve, reconstruct and repurpose the city fortress by creating an attractive destination inside as well as the outside of the walls, which would be enjoyed by both residents and tourists. The project aims to research the potential and discover the strategy model to restore new life and impose purpose to the rapidly devastating ramparts. The renewal paradigm is researched in the course of three main segments: adaptation, revitalization and repurpose.
The project aspires to preserve the existing visual identity of the city walls and raise awareness of rich cultural and historical heritage trough introducing several museums, culture facilities as well as multicultural public spaces, resulting in a prominent landmark of high visibility and the local acropolis in the revived rebranded city. Keywords: Nikšić, fortress, reconstruction, revitalization, cultural heritage, museum, public space Published in DKUM: 08.10.2021; Views: 1077; Downloads: 113 Full text (96,88 MB) |
9. Differential evolution and large-scale optimization applicationsAleš Zamuda, scientific film, scientific sound or video publication Abstract: Differential Evolution (DE) is one of the most popular, high-performance optimization algorithms with variants that have been outperforming others for years. As a result, DE has grown to accommodate wide usage for a variety of disciplines across scientific fields. Differential Evolution and Large-Scale Optimization Applications presents a research-based overview and cross-disciplinary applications of optimization algorithms. Emphasizing applications of Differential Evolution (DE) across sectors and laying the foundation for further use of DE algorithms in real-world settings, this video is an essential resource for researchers, engineers, and graduate-level students. Topics Covered : Algorithms, Optimization, Parallel Differential Evolution, Performance Improvement, Stochastic Methods, Tree Model Reconstruction. Keywords: differential Evolution, optimization, algorithms, stochastic methods, tree models, tree model reconstruction Published in DKUM: 14.05.2019; Views: 1822; Downloads: 223 Link to file This document has many files! More... |
10. Weak k-reconstruction of Cartesian productsWilfried Imrich, Blaž Zmazek, Janez Žerovnik, 2003, original scientific article Abstract: By Ulam's conjecture every finite graph ▫$G$▫ can be reconstructed from its deck of vertex deleted subgraphs. The conjecture is still open, but many special cases have been settled. In particular, one can reconstruct Cartesian products. We consider the case of ▫$k$▫-vertex deleted subgraphs of Cartesian products and prove that one can decide whether a graph ▫$H$▫ is a ▫$k$▫-vertex deleted subgraph of a Cartesian product ▫$G$▫ with at least ▫$k+1$▫ prime factors on at least ▫$k+1$▫ vertices each, and that ▫$H$▫ uniquely determines ▫$G$▫. This extends previous works of the authors and Sims. This paper also contains a counterexample to a conjecture of MacAvaney. Keywords: mathematics, graph theory, reconstruction problem, Cartesian product, composite graphs Published in DKUM: 31.03.2017; Views: 1558; Downloads: 445 Full text (197,33 KB) This document has many files! More... |