Electroencephalography (EEG) is a widely used, noninvasive, and portable neuroimaging technique with high temporal resolution. However, EEG recordings are inherently mixtures of neural signals and artifacts, making artifact separation and removal essential yet challenging. Various blind source separation (BSS) methods, particularly those based on independent component analysis (ICA), have been broadly applied to EEG but still face limitations, as artifacts often remain mixed with neural sources within the same components. To address this issue, we propose a novel framework based on independent lowrank matrix analysis (ILRMA), which captures distinct spectral structures and temporal dynamics across different source types. By unifying ILRMA's two three-dimensional outputs into a four-dimensional representation, the proposed method extends conventional component-level analysis to intra-component-level, unfolding individual components along the frequency dimension to multiple intra-components with more discriminative patterns. By this means, signal sources, including artifacts, are finely separated with less neural information loss and residual noise, which also builds a solid base for enhancing the reliability and reproducibility of downstream EEG analyses.

In the era of deep learning, training data play an essential role. Yet, their generation is expensive concerning both cost and time. Besides this, it often suffers from (1) insufficient data, (2) fixed definition of categories possibly leading to data imbalance, and (3) difficult quality control of manual annotations. In this paper, we propose an approach for automatic generation of urban building data for detection and classification from remote sensing imagery which attempts to deal with these issues. Datasets in popular format, which can be directly used for training, are created in a fully automatic pipeline from the raw data. Furthermore, the generation process can be customized to adapt the datasets to specific tasks as well as optimized according to the data characteristics of the source. We show that an automatically generated dataset can reach a comparable level to the current open datasets concerning both the quantity of instances as well as the diversity of categories. All this is achieved in a few hours without manual intervention or costs for annotation. Experiments with multiple datasets including the comparison with manually annotated data demonstrate the potential of the proposed approach.

Progress in computer vision, particularly based on machine learning, depends heavily on the availability of appropriate datasets. However, for applications like face recognition or emotion detection, this requires the collection of face images, which comprise especially privacy-sensitive biometric data. The corresponding valid ethical concerns and legal regulations regarding privacy rights limit the creation of new datasets. To reconcile the need for detailed facial image datasets with the right to privacy protection, appropriate anonymization techniques are needed. To this end, we suggest a pipeline to repurpose a face swapping tool for de-identification by combining it with synthetic image generation and a novel procedure to select source images to improve the trade-off between data utility retention and privacy enhancement. A quantitative comparison of our results to other de-identification approaches shows that our method leads to better retention of facial expressions while providing adequate privacy protection. Thus, applying this procedure to face image datasets before publication could help mitigate privacy concerns.

One of the common tasks in remote sensing research is object recognition, i.e., detection followed by classification, in satellite images. Object recognition algorithms are widely used in various industrial and military applications. We contribute to the academic research by focusing on airplane recognition in satellite images. In our paper, we utilize existing state-of-the-art deep learning models combined with (1) separating detection and classification and (2) normalizing the input image data concerning rotation in a way that the head of an aircraft always points in the same given direction. This leads to a considerable improvement of both detection and classification accuracy of airplanes as demonstrated by the results from the FAIR1M data set.

Training data is an essential ingredient for the development of deep learning approaches. Yet, the preparation of training datasets for building detection and classification in remote sensing images implies substantial manual work and is, therefore, expensive concerning both labor charges and time. Since manual annotation also strongly depends on the experience and expertise of the annotators, quality control is an unavoidable issue. It is, thus, of great interest to explore means to reduce the manual part of dataset generation while keeping the quality of the annotation at an acceptable level.In this paper, we present a novel approach to creating training datasets for individual building detection and classification from remote sensing imagery consisting of a fully automatic pipeline. Using 3D city models and high-resolution imagery as input, annotations including building footprint and their attributes are automatically generated and combined with the corresponding image segments into a standard dataset complying with the COCO format. Experiments comprising also the comparison to manually labeled datasets demonstrate the potential of the proposed work.

Datasets play a key role in developing superior building detection approaches. However, most of the previous work focuses on accurate building masks and scale expansion, while the categories are always missing, which hinders the further analysis of urban development and cultures. Therefore, we propose a benchmark for building detection and fine-grained classification from very high-resolution (VHR) satellite imagery. An extensive annotation is performed for about 0.5 million building instances with 12 fine-grained roof types and individual polygons. The annotation of building functions of two cities in the previous version (UBCv1) [1] is also integrated. To ensure the building variety, it consists of VHR optical images of 20 unique cities worldwide with various landforms and styles of architecture. Its variety and fine-grained categories pose great challenges and meanwhile provide a foundation for the building extraction and fine-grained classification on a global scale. Besides, 17 cities are provided with finely aligned Synthetic Aperture Radar (SAR) images, which can be employed for the development and evaluation of approaches optionally based on optical, SAR, or multi-modal images. Significantly, the proposed benchmark is used as the base of the 2023 IEEE GRSS Data Fusion Contest [2]. The dataset and codes of the baseline methods are available at: https://github.com/AICyberTeam/UBC-dataset/tree/UBCv2.

In this paper, a novel algorithm for the automatic online improvement of the extrinsic camera parameters of a stereo image pair is introduced. To this end, the well-known dense stereo matching method PatchMatch stereo (PM) is extended for the pixelwise estimation of a discrepancy between the expected epipolar line and the actual correspondence. The availability of an initial guess of the camera parameters is assumed. Next, the estimated disparity map is filtered for highly stable and accurate correspondences that cover preferably the complete image. For this reason, we extend a quality estimation method adapted to Semi-Global Matching (SGM) derived disparity maps for general disparity maps. Finally, the set of stable and accurate correspondences from the disparity map is used for the estimation of the extrinsic camera parameters by means of the five-point algorithm in a RANSAC (random sample consensus) framework. Our algorithm can estimate optimized disparity maps and is able to adjust for errors in the relative camera pose. It can even correct epipolar errors of tens of pixels in highresolution images. We demonstrate that the proposed algorithm allows for robust and accurate estimation of the extrinsic camera parameters on datasets that provide weaklycalibrated image pairs.

We present a method for Modeling Urban Scenes from Pointclouds (MUSP). In contrast to existing approaches, MUSP is robust, scalable and provides a more complete description by not making a Manhattan-World assumption and modeling both buildings (with polyhedra) as well as the non-planar ground (using NURBS). First, we segment the scene into consistent patches using a divide-and-conquer based algorithm within a nonparametric Bayesian framework (stick-breaking construction). These patches often correspond to meaningful structures, such as the ground, facades, roofs and roof superstructures. We use polygon sweeping to fit predefined templates for buildings, and for the ground, a NURBS surface is fit and uniformly tessellated. Finally, we apply boolean operations to the polygons for buildings, buildings parts and the tesselated ground to clip unnecessary geometry (e.g., facades protrusions below the non-planar ground), leading to the final model. The explicit Bayesian formulation of scene segmentation makes our approach suitable for challenging datasets with varying amounts of noise, outliers, and point density. We demonstrate the robustness of MUSP on 3D pointclouds from image matching as well as LiDAR.

In this paper we present a robust and efficient approach for automatic urban scene classification based on imagery and elevation data. Scene classification is of great interest for a broad spectrum of applications, e.g., city models, urban planning and land cover/use. Because of the availability of high resolution imagery and the corresponding scene complexity as well as heterogeneous appearance of objects, scene classification of urban areas is still challenging with respect to accuracy and efficiency. To this end, we propose "relative features", which are intra-class stable and inter-class discriminative, instead of absolute ones for color and geometry to deal with object diversity and scene complexity. The proposed approach provides a pixel-wise as well as a patch-wise schemes with (1) robustness against the variability of object appearance, (2) adaptation to undulating terrain and (3) fully-parallel processing for feature extraction and classification. Experiments on public benchmark and self-acquired data demonstrate the potential of the proposed approach.

Local fusion of disparity maps allows fast parallel 3D modeling of large scenes that do not fit into main memory. While existing methods assume a constant disparity uncertainty, disparity errors typically vary spatially from tenths of pixels to several pixels. In this paper we propose a method that employs a set of Gaussians for different disparity classes, instead of a single error model with only one variance. The set of Gaussians is learned from the difference between generated disparity maps and ground-truth disparities. Pixels are assigned particular disparity classes based on a Total Variation (TV) feature measuring the local oscillation behavior of the 2D disparity map. This feature captures uncertainty caused for instance by lack of texture or fron to-parallel bias of the stereo method. Experimental results on several datasets in varying configurations demonstrate that our method yields improved performance both qualitatively and quantitatively.

The coregistration of complex image pairs is a very important step in synthetic aperture radar interferometry (InSAR) data processing. This letter proposes a novel coregistration method that only needs three arbitrary parts of the two complex images instead of four parts in the existing coregistration methods. This method constitutes an integrated three-part method for InSAR data processing with our contoured-correlation-interferometry method for phase-image generation. Saving one part transmission makes a significant advantage when processing SAR images on satellites. Furthermore, we demonstrate that, by means of using fringe contoured windows instead of squared windows, the accuracy of the coregistration for both the three-part coregistration method and the existing methods can be improved considerably

This paper presents a means for automatic generation of a level of detail (LOD) representation of three‐dimensional (3D) building data, with the formally well‐defined scale‐spaces as the underlying theory. More specifically, we propose an approach that employs vector‐based mathematical morphology and discrete curvature‐space to generalize two‐dimensional (2D) building outlines as well as 3D building surfaces. Scale‐space events that are related to the semantics of objects are the major triggers of the generalization. The implemented approach preserves right angles. Additionally, more heuristic means to square buildings are presented. The test results have validated the approach.

Due to the speckle effect of coherent imaging the detection of lines in SAR scenes is considerably move difficult than in optical images. A new approach to detect lines in noisy images using a Markov random field (MRF) model and Bayesian classification is proposed. The unobservable object classes of single pixels are assumed to fulfil the Markov condition, i.e. to depend on the object classes of neighboring pixels only. The influence of neighboring line pixels is formulated based on potentials derived from a random walk model. Locally, the image data is evaluated with a rotating template. As SAR intensity data is deteriorated by multiplicative noise, the response of the local line detector is a normalized intensity ratio which results in a constant false alarm rate. The approach integrates intensity, coherence from interferometric processing of a SAR scene pair, and given Geographic Information System (GIS) data.

The paper presents a system for automatic extraction of high level information from land register maps, e.g. parcels, buildings or roads. The system uses explicit knowledge of the map, which is given by the legend of the map, the drawing rules, and the constraints by functionality. The knowledge is represented by frames and semantic networks. The system uses four levels of representation and processing and a mixed strategy by feedback cycles between the levels. Results are presented for extraction of legal information (parcels and boundary stones) and topographic information (buildings, roads, sidewalls, farmland) from land register maps of scale 1:1000 and 1:5000. The precision of the extracted legal information is sufficient for the given task.

Data acquisition is the bottleneck for the introduction of geographic information systems (GIS). This paper presents a system for automatic extraction of semantic information from land register maps. The system uses explicit knowledge of the map, which is given by the legend of the map, the drawing rules, and the objects functionality. The knowledge is represented by frames and semantic networks. The system uses four levels of representation and processing: 1. raster image, 2. image graph of lines and junctions, 3. graphics and text objects, and 4. semantic objects of the map. The interpretation of the map is performed by instantation of the concept nodes of the semantic network. Results are presented for extraction of legal information (parcels and boundary stones) and topographic information (buildings, roads, sidewalks, farmland) from land register maps of scale 1:1000 and 1:5000.