Multiview Stereo (MVS) reconstruction techniques have made significant advancements with the development of deep learning. However, their performance often deteriorates in low-light conditions, where feature extraction and matching become challenging. Traditional image enhancement solutions are insufficient for MVS tasks in low illumination, relying on manual adjustments. We introduce an end-to-end MVS framework incorporating a diffusion-based image enhancement algorithm with MVS to build an end-to-end framework for improving the performance of MVS in low-light conditions. This integration improves color rendering and visualization of 3D reconstructions and slightly enhances geometric shapes. Our method uses a feature adapter to integrate the enhanced images from the Low-light Diffusion model into CasMVSNet, refining the feature maps in poorly lit environments. Validation on the DTU and Tanks and Temples datasets demonstrates our model’s robustness and generalizability across various lighting conditions and MVS pipelines, including GeoMVSNet and MVSNet. Our approach simplifies the training process by requiring only the training of an adapter rather than a multi-view image enhancement model, underscoring the effectiveness of incorporating image enhancement into learning-based MVS frameworks for low-light conditions.

This project in Rotterdam, centered on the Meent, represents a new beginning for revitalizing the city with a Symbiocene mindset through architectural design. As urban populations grow, biodiversity declines and technology advances exponentially, it's imperative to address how we interact with our environment. The Manifesto advocates for eight key relationships to usher in the Symbiocene era: celebrate, master, connect, shelter, concert, make, recede, and familiarize.

The architectural design project aims to foster these relationships within a residential setting by renovating existing post-war buildings into a new kind of living space. The goal is to create an environment where residents can experience and cultivate these relationships through their daily lives.

The design includes various apartment typologies, including cohousing, within the renovated post-war block. The building is transformed into a canopy-like structure by removing internal boundaries and rewilding the courtyard, encouraging interaction with the living environment. Public access from the Meent side invites passersby to engage with the space, while the open grid scaffolding at the back provides a dynamic area where humans, non-humans, and more-than-humans can coexist and learn from one another.

Transitioning to the eight relationships detailed in the Manifesto:

Celebrate – Recognize and appreciate the gifts of our environment, inspired by Robin Wall Kimmerer's concept of the Earth's gift economy. It calls for a celebration of Gaia, fostering gratitude and reverence for nature.
Master – Continuous learning and the pursuit of knowledge are crucial for understanding our environment. This relationship emphasizes an ongoing quest for understanding and mastery of our world.
Connect – Engage with the environment through sensory experiences and active participation to foster respect and responsibility towards nature.
Shelter – Balance protection and security with interaction with the natural world, avoiding isolation from the environment.
Concert – Collaborate with other species for a balanced ecosystem, understanding and respecting various forms of interaction within our ecosystem.
Make – Focus on the ethical implications of human creativity and design, fostering positive behavior and aligning with symbiotic dreams.
Recede – Step back and allow nature to reclaim spaces, appreciating and protecting neglected and abandoned areas for biodiversity and ecological balance.
Familiarize – Understand emerging technologies as new "species" with distinct behaviors and impacts, thoughtfully integrating technology to enhance our connection with the natural world.

As summarized, the project aims to create a living environment that not only houses people but also nurtures these eight relationships, fostering a deeper connection between residents and their surroundings. By reimagining the urban landscape in this way, we can move towards a more sustainable and harmonious future.

We explored the possibility of improving cross-view matching performance with self-supervised learning techniques and perform interpretations in terms of the embedding space of image features. The effect of pre-training by contrastive learning is verified quantitatively by experiments, and also exhibited by visualization of the feature space.

Roads in modern cities facilitate different types of users, including car drivers, cyclists, and pedestrians. These different users often have a designated section of the road to operate on. Road management, e.g., by municipalities, needs to take this sectioning into account, preferably in an efficient way. Mobile laser scanning (MLS) point clouds provide accurate and dense three-dimensional (3D) measurements of road scenes, showing strong mapping capabilities, although their massive data volume and lack of structure still bring difficulties in automatic processing. Methods for the automatic classification of road surface types are still largely lacking, and the existing methodology did not consider the potential of MLS point clouds yet. In recent years, point cloud understanding through deep neural networks has achieved breakthroughs. However, perceiving large-scale point clouds by deep learning depends on aggregating local features and progressive downsampling, to extract rich contextual information. As a consequence, low-level features that reveal details in point clouds may not be well preserved, possibly resulting in ambiguous delineation in point cloud classification. For road mapping, inaccurate classification of points on road boundaries hinders the generation of high-quality map products. Some existing deep learning methods propose to mitigate the fuzzy classification near boundaries, either by utilizing refinement for network predictions, or by indirectly modifying neighboring weights when summarizing local information. Approaches to achieving a satisfying overall performance, while maintaining accurate delineation, still need investigation. In this study, we propose a novel approach for road type classification of MLS point clouds in dense urban areas based on a deep neural network. We follow the main architecture of RandLA-Net, a point-wise neural network designed for large-scale point cloud processing. To alleviate the ambiguous delineation of point cloud classification, we propose two strategies. The first is to refine predictions of RandLA-Net by conditional random field. In the second strategy, we incorporate boundary constraints in the network by introducing a novel distance label for each road surface point to represent the distance to its closest boundary. The distance prediction task is combined with road type classification by adding another branch in RandLA-Net to formulate multi-task learning. Through experiments, we show that 3D point cloud semantic segmentation by deep learning is applicable for road type classification. Also, the multi-task learning strategy is verified to be more effective in improving the delineation performance. Using MLS point clouds acquired from 5 German cities (Hamburg, Delmenhorst, Bremerhaven, Hannover, and Oldenburg), we classify road points separately into different usages (sidewalk, cycling path, rail track, parking area, motorway, green area, and island without traffic) and materials (cobblestone, asphalt, plates, unpaved, and railway). When adopting Hannover and Oldenburg for testing, and the other three cities for training, we obtain a mean intersection over union of 46.1% for usage type and 52.0% for material type with the multi-task learning strategy and input features (x, y, z, R, G, B, intensity), outperforming the original RandLA-Net by approximately 4%. Moreover, from the point cloud classification results, we achieve lightweight polygon representations of road objects in different types through post-processing, which is demonstrated to perform better than an image semantic segmentation-based solution quantitatively and qualitatively.

An increasing load on the Dutch rail network requires new solutions for railway monitoring as increasing wear and tighter schedules limit the possibilities of regular maintenance. UAV (Unmanned Aerial Vehicle) based photogrammetry is such a solution which limits the presence of people near the railway and train hindrance. Photogrammetric feature extraction methods struggle with extracting features from the rusty rail and reflective roll band. Therefore line features are used to aid reconstruction. Based on SfM (Structure-from-Motion) camera parameters, a line-based reconstruction is created of two data-sets. a A line detection algorithm is used to find line segments in the images. Found segments are matched between pairs of images by pixel window cross-correlation and geometric boundaries. Matched segments are then compared to find sets of three matching images. Reconstruction is done by minimizing line reprojection error in an iterative least squares solution. Results show promise for manually matched segments, however unfavourable automatic matching results prevent large scale reconstruction. A case study reveals matching struggles with camera rotation between images, matching problem mitigation leaves much room for improvement.

Semantic segmentation of aerial point clouds with high accuracy is significant for many geographical applications, but is not trivial since the data is massive and unstructured. In the past few years, deep learning approaches designed for 3D point cloud data have made great progress. Pointwise neural networks, such as PointNet and its extensions, show their ability to process 3D point clouds, especially in classification and semantic segmentation. In this work, we implement DGCNN (Dynamic Graph CNN), which combines PointNet with Graph CNN, and extend its semantic segmentation application from indoor scenes to an aerial point cloud dataset: The Current Elevation File Netherlands (AHN), which was produced by airborne laser scanners for the whole Netherlands. Point clouds from the iteration AHN3 are classified into four classes: ground, building, water and others (including vegetation, railways, etc). Moreover, DGCNN splits the input point cloud into regular blocks before operating on it and processes each block independently, which limits the effective range (receptive field) of the network to some extent. Thus, the second aim of this work is to investigate the impact of the effective range on the performance of DGCNN by adjusting two crucial parameters: the block size and the neighborhood size k in k-NN graphs. It turns out that enlarging the block size or k helps to improve the overall accuracy of DGCNN, but cannot ensure better segmentation results from each individual class. With the block size 50 m and k=20, the most balanced F1 scores for all classes and an overall accuracy of 93.28% are achieved. Based on the evaluation for each setting with a certain block size and k, we also manage to further improve the overall accuracy to 93.51% by combining smaller-scale (with block size 30 m) and larger-scale (with block size 50 m) segmentation results, with k=20.

Trees are an important aspect of the world around us, and play a sufficient role in our daily lives. They contribute to human health and well-being in various ways. Tree inventory and monitoring are of great interest for biomass estimations and changes in the purifying effect on the air. It is a very time consuming and cost inefficient way to check every tree in and around a city or town, therefore there is further research required in the use of AHN data. Together with the “tree information data set” formthemunicipality ofDelft, the location and the corresponding point cloud of tree different species of trees are selected. For the species of interest, Aesculus Hippocastanum, Acer Saccharinum and Platanus x Hispanica, different characteristics are determined. In this research six different characteristics are estimated; Height, Trunk Height, Normalized Trunk Height, Canopy Projected Area, Normalized Canopy Projected Area, Ratio of Diameters, Normalized Ratio of Diameter, Centre of Gravity and at least the Normalized Centre of Gravity. These characteristics are used as features for the Random Forest Classification, Consequently the Confusion Matrix is used as performance measurement. The results of a test of 30 pointclouds, per species of interest, show that the Random Forest Classification is able to classify individual trees. However, these three different species cannot by sufficiently classified using clustering.

A cross-domain visual place recognition (VPR) task is proposed in this work, i.e., matching images of the same architectures depicted in different domains. VPR is commonly treated as an image retrieval task, where a query image from an unknown location is matched with relevant instances from geo-tagged gallery database. Different from conventional VPR settings where the query images and gallery images come from the same domain, we propose a more common but challenging setup where the query images are collected under a new unseen condition. The two domains involved in this work are contemporary street view images of Amsterdam from the Mapillary dataset (source domain) and historical images of the same city from Beeldbank dataset (target domain). We tailored an age-invariant feature learning CNN that can focus on domain invariant objects and learn to match images based on a weakly supervised ranking loss. We propose an attention aggregation module that is robust to domain discrepancy between the train and the test data. Further, a multi-kernel maximum mean discrepancy (MK-MMD) domain adaptation loss is adopted to improve the cross-domain ranking performance. Both attention and adaptation modules are unsupervised while the ranking loss uses weak supervision. Visual inspection shows that the attention module focuses on built forms while the dramatically changing environment are less weighed. Our proposed CNN achieves state of the art results (99% accuracy) on the single-domain VPR task and 20\% accuracy at its best on the cross-domain VPR task, revealing the difficulty of age-invariant VPR.

Climate change has been a heated topic in recent years, and the mass loss of ice sheets is one aspect of it. The Antarctic Ice Sheet has experienced certain mass loss in the form of ice-shelf collapse and (sub-)surface melting, but thorough study remains limited due to the remote location of the continent. Therefore, remote sensing is expected to provide valuable information on Antarctica, in order to monitor its mass balance and gain insights on the extent of climate change.

As one of the factors to the mass loss in Antarctica, subsurface melt can be critical yet hard to capture. The limitation of remote sensing data, especially valid optical images over polar regions may add to the issue. This study aims to exploit the potentiality of SAR Interferometry in detecting subsurface melt, as the microwave bands are not affected by weather and illumination, and the interferometric data may provide supportive information reflecting the properties of the physical environment. It is expected that by using this technique, the gap can be filled in when optical images are not available, or pure SAR images are not informative. The technique is applied to two ice shelves in East Antarctica, Roi Baudouin Ice Shelf and Amery Ice Shelf. And this study is expected to be operated over a broader scale such as the Antarctic continent and Greenland.