Research - Shape Analysis and Geometry Processing
-
Shape Classification of Building Information Models using Neural Networks
I. Evangelou, N. Vitsas, G. Papaioannou, M. Georgioudakis, A. Chatzisymeon, short paper, Proc. Eurographics Workshop on 3D Object Retrieval (3DOR), 2021.
Abstract. The Building Information Modelling (BIM) procedure introduces specifications and data exchange formats widely used by the construction industry to describe functional and geometric elements of building structures in the design, planning, cost estimation and construction phases of large civil engineering projects. In this paper we explain how to apply a modern, low-parameter, neural-network-based classification solution to the automatic geometric BIM element labeling, which is becoming an increasingly important task in software solutions for the construction industry. The network is designed so that it extracts features regarding general shape, scale and aspect ratio of each BIM element and be extremely fast during training and prediction. We evaluate our network architecture on a real BIM dataset and showcase accuracy that is difficult to achieve with a generic 3D shape classification network.
Downloads: the paper
Implementation repository: https://github.com/cgaueb/deep_bim
-
Fast Radius Search Exploiting Ray Tracing Frameworks
I. Evangelou, G. Papaioannou, K. Vardis, A. A. Vasilakis, Journal of Computer Graphics Techniques (JCGT), vol. 10, no. 1, 25-48, 2021.
Abstract. Spatial queries to infer information from the neighborhood of a set of points are very frequently performed in rendering and geometry processing algorithms. Traditionally, these are accomplished using radius and k-nearest neighbors search operations, which utilize kd-trees and other specialized spatial data structures that fall short of delivering high performance. Recently, advances in ray tracing performance, with respect to both acceleration data structure construction and ray traversal times, have resulted in a wide adoption of the ray tracing paradigm for graphics-related tasks that spread beyond typical image synthesis. In this work, we propose an alternative formulation of the radius search operation that maps the problem to the ray tracing paradigm, in order to take advantage of the available GPU-accelerated solutions for it. We demonstrate the performance gain relative to traditional spatial search methods, especially on dynamically updated sample sets, using two representative applications: geometry processing point-wise operations on scanned point clouds and global illumination via progressive photon mapping.
Online paper: http://jcgt.org/published/0010/01/02/
-
PU learning-based recognition of structural elements in architectural floor plans
I. Evangelou, M. Savelonas, G. Papaioannou, Multimedia Tools & Applications, 2021. https://doi.org/10.1007/s11042-020-10295-9.
Abstract. This work introduces a computational method for the recognition of structural elements in architectural floor plans. The proposed method requires minimal user interaction and is capable of effectively analysing floor plans in order to identify different types of structural elements in various notation styles. It employs feature extraction based on Haar kernels and PU learning, in order to retrieve image regions, which are similar to a user-defined query. Most importantly, apart from this user-defined query, the proposed method is not dependent on learning from labelled samples. Therefore, there is no need for laborious annotations to form large datasets in various notation styles. The experimental evaluation has been performed on a publicly available and diverse dataset of floor plans. The results show that the proposed method outperforms a state-of-the-art method, with respect to retrieval accuracy. Further experiments on additional floor plans of various notation styles, demonstrate its general applicability.
-
Exploiting Unbroken Surface Congruity for the Acceleration of Fragment Reassembly
M. A. Savelonas, A. Andreadis, G. Papaioannou, P. Mavridis, Proc. EUROGRAPHICS Workshop on Graphics and Cultural Heritage, 2017.
Abstract. Virtual reassembly problems are often encountered in the cultural heritage domain. The reassembly or "puzzling" problem is typically described as the process for the identification of corresponding pieces within a part collection, followed by the clustering and pose estimation of multiple parts that result in a virtual representation of assembled objects. This work addresses this problem with an efficient, user-guided computational approach. The proposed approach augments the typical reassembly pipeline with a smart culling step, where geometrically incompatible fragment combinations can be quickly rejected. After splitting each fragment into potentially fractured and intact facets, each intact facet is examined for prominent linear or curved structures and a heuristic test is employed to evaluate the plausibility of facet pairs, by comparing the number of feature curves associated with each facet, as well as the geometric texture of associated intact surfaces. This test excludes many pairwise combinations from the remaining part of the reassembly process, significantly reducing overall time cost. For all facet pairs that pass the initial plausibility test, pairwise registration driven by enhanced simulated annealing is applied, followed by multipart registration. The proposed reassembly approach is evaluated on real scanned data and our experiments demonstrate an increase in efficiency that ranges from 30% to more than 500% in some cases, depending on the number of culled combinations.
-
Predictive Digitisation of Cultural Heritage Objects
I. Pratikakis, M. Savelonas, P. Mavridis, G. Papaioannou, K. Sfikas, F. Arnaoutoglou, D. Rieke-Zapp, Multimedia Tools and Applications, Springer, pp. 1-31, June, 2017.
Abstract. 3D digitisation has been instrumental in the cultural heritage domain for over a decade, contributing to the digital preservation and dissemination of cultural heritage. Still, the typical 3D acquisition workflow remains complex and time-consuming. This work presents the concept of predictive digitisation by means of a platform, aiming to speed-up and simplify 3D digitisation, exploiting similarities in digital repositories of Cultural Heritage objects.
-
From Reassembly to Object Completion: A Complete Systems Pipeline
G. Papaioannou, T. Schreck, A. Andreadis, P. Mavridis, R. Gregor, I. Sipiran, K. Vardis, ACM Journal on Computing and Cultural Heritage, 10(2), Article 8, 2017.
Abstract. The problem of restoration of broken artefacts, where large parts could be missing, is of high importance in archaeology. The typical manual restoration can become a tedious and error-prone process, which also does not scale well. In recent years, many methods have been proposed for assisting the process, most of which target specialized object types or operate under very strict constraints. We propose a digital shape restoration pipeline consisting of proven, robust methods for automatic fragment reassembly and shape completion of generic three-dimensional objects of arbitrary type. In this pipeline, first we introduce a novel unified approach for handling the reassembly of objects from heavily damaged fragments by exploiting both fracture surfaces and salient features on the intact sides of fragments, when available. Second, we propose an object completion procedure based on generalized symmetries and a complementary part extraction process that is suitable for driving the fabrication of missing geometry. We demonstrate the effectiveness of our approach using real-world fractured objects and software implemented as part of the EU-funded PRESIOUS project, which is also available for download from the project site.
Downloads: Official paper (via ACM Authorizer service)
-
GPU Accelerated Computation of Geometric Descriptors in Parametric Space
A. Andeadis, G. Papaioannou, P. Mavridis, chapter, in book Computer Vision, Imaging and Computer Graphics Theory and Applications, Springer, pp 41-61, 2016.
Abstract. We present a novel generic method for the fast and accurate computation of geometric descriptors. While most existing approaches perform the computations directly on the geometric representation of the model, our method operates in parametric space, decoupling the computational complexity from the underlying mesh geometry. In contrast to other parametric space approaches, our method is not restricted to specic descriptors or parameterisations of the surface. By using the parametric space representation of the mesh geometry, we can trivially exploit massive parallel GPU architectures and achieve interactive computation times, while maintaining high accuracy. This renders the method suitable for computations involving large areas of support and animated shapes.
Downloads: Author-prepared version of the paper
-
Object Completion using k-Sparse Optimization
P. Mavridis, I. Sipiran, A. Andreadis, G. Papaioannou, Computer Graphics Forum (proc. Pacific Graphics), vol. 34, no. 7, 2015.
Abstract. We present a new method for the completion of partial globally-symmetric 3D objects, based on the detection of partial and approximate symmetries in the incomplete input dataset. In our approach, symmetry detection is formulated as a constrained sparsity maximization problem, which is solved efficiently using a robust RANSAC-based optimizer. The detected partial symmetries are then reused iteratively, in order to complete the missing parts of the object. A global error relaxation method minimizes the accumulated alignment errors and a non-rigid registration approach applies local deformations in order to properly handle approximate symmetry. Unlike previous approaches, our method does not rely on the computation of features, it uniformly handles translational, rotational and reflectional symmetries and can provide plausible object completion results, even on challenging cases, where more than half of the target object is missing. We demonstrate our algorithm in the completion of 3D scans with varying levels of partiality and we show the applicability of our approach in the repair and completion of heavily eroded or incomplete cultural heritage objects.
Downloads: Author-prepared version of the paper
-
Generalized Digital Reassembly using Geometric Registration
A. Andreadis, G. Papaioannou, P. Mavridis, Proc. Digital Heritage 2015.
Abstract. We present a novel and generic user-guided approach for the digital reconstruction of cultural heritage finds from fragments, which operates directly on generic 3D objects. Central to our approach is a three-tier geometric registration approach that addresses the reassembly problem using i) the contact surface of the fractured objects, ii) feature curves on the intact surfaces and iii) partial object symmetries. In contrast to most existing methodologies, our approach is more generic and addresses even the most difficult cases, where contact surface is unusable, small or absent. We evaluate our method using digitized fragments from the Nidaros Cathedral.
Downloads: Author-prepared version of the paper
-
Fractured 3D Object Restoration and Completion
A. Andreadis, R. Gregor, I. Sipiran, P. Mavridis, G. Papaioannou, T. Schreck, ACM SIGGRAPH 2015 Poster.
Abstract. The problem of object restoration from eroded fragments, where large parts could be missing, is of high relevance in archaeology. Manual restoration is possible and common in practice but it is a tedious and error-prone process, which does not scale well. Solutions for specific parts of the problem have been proposed but a complete reassembly and repair pipeline is absent from the bibliography. We propose a shape restoration pipeline consisting of appropriate methods for automatic fragment reassembly and shape completion. We demonstrate the effectiveness of our approach using real-world fractured objects. We suggest two main restoration phases. In the first phase, fragments are reassembled. The reassembly solution results from finding the Minimum Spanning Forest for pairwise matches between fragment contact surfaces, which have been identified in a preprocessing step . The reassembly is guided by global error relaxation and can also make use of external feature curves on the fragments. The obtained reassembly solution typically misses some parts of the shape, due to missing or eroded fragments. Therefore, in the second phase we compute plausible complete versions of the reassembles partial shapes. This is done by robust detection of global shape symmetries which relies on local shape features. Completion of non-symmetric shapes is assisted by template repair shapes retrieved by a partial 3D similarity search. The final shape is finished by merging and smoothing of the obtained parts, inpainting of missing local shape information, and export of synthesized missing parts for physical restoration.
Downloads: one-page abstract short video one-page poster
-
Efficient Sparse ICP
P. Mavridis, A. Andreadis, G. Papaioannou, Computer Aided Geometric Design (Proc. Geometric Modeling and Processing), vol. 35–36, pp. 16-26, 2015.
Abstract. The registration of two geometric surfaces is typically addressed using variants of the Iterative Closest Point (ICP) algorithm. The Sparse ICP method formulates the problem using sparsity-inducing norms, significantly improving the resilience of the registration process to large amounts of noise and outliers, but introduces a significant performance degradation. In this paper we first identify the reasons for this performance degradation and propose a hybrid optimization system that combines a Simulated Annealing search along with the standard Sparse ICP, in order to solve the underlying optimization problem more efficiently. We also provide several insights on how to further improve the overall efficiency by using a combination of approximate distance queries, parallel execution and uniform subsampling. The resulting method provides cumulative performance gain of more than one order of magnitude, as demonstrated through the registration of partially overlapping scans with various degrees of noise and outliers.
Downloads: author-prepared version of the paper
-
Fractured Object Reassembly via Robust Surface Registration
P. Mavridis, A. Andreadis, G. Papaioannou, proc. Eurographics Conference (short paper). 2015.
Abstract. The reassembly of fractured 3D objects from their parts is an important problem in cultural heritage and other domains. We approach reassembly from a geometric matching perspective and propose a pipeline for the automatic solution of the problem, where an efficient and generic three-level coarse-to-fine search strategy is used for the underlying global optimization. Key to the efficiency of our approach is the use of a discretized approximation of the surfaces’ distance field, which significantly reduces the cost of distance queries and allows our method to systematically search the global parameter space with minimal cost. The resulting reassembly pipeline provides highly reliable alignment, as demonstrated through the reassembly of fractured objects from their fragments and the reconstruction of 3D objects from partial scans, showcasing the wide applicability of our methodology.
Downloads: author-prepared version of the paper paper video 3D reassembly models
-
A Parametric Space Approach to the Computation of Multi-Scale Geometric Features
A. Andreadis, G. Papaioannou, P. Mavridis, proc. GRAPP 2015.
Abstract. In this paper we present a novel generic method for the fast computation of geometric features at multiple scales, on arbitrarily complex models, which operates in the parametric space. The majority of the existing methods compute local features directly on the geometric representation of the model. Our approach decouples the computational complexity from the underlying geometry and in contrast to other parametric spacemethods, it is not restricted to a specific feature or parameterization of the surface. We showcase that the method performs accurately and at interactive rates, even for large feature areas of support, rendering the method suitable for animated shapes.
Downloads: author-prepared version of the paper paper video
-
Towards Automated 3D Reconstruction of Defective Cultural Heritage Objects
R. Gregor, I. Sipiran, G. Papaioannou, T. Schreck, A. Andreadis, P. Mavridis, Proc. Eurographics Workshop on Graphics and Cultural Heritage, pp. 135-144, 2014
Abstract. Due to recent improvements in 3D acquisition and shape processing technology, the digitization of Cultural Heritage (CH) artifacts is gaining increased application in context of archival and archaeological research. This increasing availability of acquisition technologies also implies a need for intelligent processing methods that can cope with imperfect object scans. Specifically for Cultural Heritage objects, besides imperfections given by the digitization process, also the original artifact objects may be imperfect due to deterioration or fragmentation processes. Currently, the reconstruction of previously digitized CH artifacts is mostly performed manually by expert users reassembling fragment parts and completing imperfect objects by modeling. However, more automatic methods for CH object repair and completion are needed to cope with increasingly large data becoming available. In this conceptual paper, we first provide a brief survey of typical imperfections in CH artifact scan data and in turn motivate the need for respective repair methods. We survey and classify a selection of existing reconstruction methods with respect to their applicability for CH objects, and then discuss how these approaches can be extended and combined to address various types of physical defects that are encountered in CH artifacts by proposing a flexible repair workflow for 3D digitizations of CH objects. The workflow accommodates an automatic reassembly step which can deal with fragmented input data. It also includes the similarity-based retrieval of appropriate complementary object data which is used to repair local and global object defects. Finally, we discuss options for evaluation of the effectiveness of such a CH repair workflow.
Downloads: author-prepared version of the paper
-
Facet Extraction and Classification for the Reassembly of Fractured 3D Objects
A. Andreadis, P. Mavridis, G. Papaioannou, Eurographics poster. Strasbourg, 2014.
Abstract. The reassembly of fractured 3D objects is a critical problem in computational archaeology, and other application domains. An essential part of this problem is to distinguish the regions of the object that belong to the original surface from the fractured ones. A general strategy to solve this region classification problem is to first divide the surface of the object into distinct facets and then classify each one of them based on statistical properties. While many relevant algorithms have been previously proposed, a comparative evaluation of some well-known segmentation strategies, when used in the context of such a problem, is absent from the bibliography. In this poster we present our ongoing work on the evaluation of the performance and quality of segmentation algorithms when operating on fractured objects. We also present a novel method for the classification of the segmented regions to intact and fractured, based on their statistical properties.
Downloads: Poster summary A4 version of the poster
Reference: BibTex
-
3D Object Repair Using 2D Algorithms
P. Stavrou, P. Mavridis, G. Papaioannou, G. Passalis, T. Theoharis, Proc. 5th International Workshop on Computer Graphics and Geometric Modeling (CGGM 2006), Lecture Notes in Computer Science, 2006, Volume 3992, Computational Science – ICCS 2006, pp. 271-278.
Abstract. A number of three-dimensional algorithms have been proposed to solve the problem of patching surfaces to rectify and extrapolate missing information due to model problems or bad geometry visibility during data capture. On the other hand, a number of similar yet more simple and robust techniques apply to 2D image data and are used for texture restoration. In this paper we make an attempt to bring these two-dimensional techniques to the 3D domain due to their obvious advantage of simplicity and controllability. Creating a depth image with the help of a voxelisation algorithm will allow us to apply a variety of image repair algorithms in order to mend a 3D object. The use of three variations of the texture synthesis algorithm is investigated. Constrained texture synthesis and its variations using the Haar wavelet and image decomposition methods are also proposed in order to preserve patterns appearing on the object while trying to maintain its geometry intact.
Downloads: the paper
Reference: BibTex
-
Efficient 3D Object Retrieval Using Depth Images
N. Vajramushti, I. A. Kakadiaris, T. Theoharis, G. Papaioannou, 6th ACM SIGMM International Workshop on Multimedia Information Retrieval (ACM SIGMM MIR '04), New York, USA, pp. 189-196, 2004.
Abstract. In this paper, we present a new three-dimensional object retrieval method. This method employs depth buffers for representing and comparing the objects. Specifically, multiple depth buffers per object (computed from different points of view) are compared for surface and volume similarity. Our method is easily extensible for hierarchical comparisons at multiple resolutions and is highly parallelizable. We have employed this method for both inter-class and intra-class retrieval tasks on a gallery of over 3 000 three-dimensional objects of vehicles with very encouraging results. The accuracy of the method depends on the number of depth buffers and the depth buffer resolution.
Downloads: the paper
Reference: BibTex
-
On the Automatic Assemblage of Arbitrary Broken Solid Artefacts
G. Papaioannou, E. A. Karabassi, Image and Vision Computing, Elsevier, 21(5), pp. 401-412, 2003.
Abstract. Presented here is a fast method that combines curve matching techniques with a surface matching algorithm to estimate the positioning and respective matching error for the joining of three-dimensional fragmented objects. Furthermore, this paper describes how multiple joints are evaluated and how the broken artefacts are clustered and transformed to form potential solutions of the assemblage problem.
Downloads: the paper
Reference: BibTex doi:10.1016/S0262-8856(03)00008-8
-
Fast Fragment Assemblage Using Boundary Line and Surface Matching
G. Papaioannou, T. Theoharis, Proc. IEEE/CVPR Workshop on Applications of Computer Vision in Archaeology (ACVA), 2003.
Abstract. In the recent past, fragment matching has been treated in two different approaches, one using curve matching methods and one that compares whole surfaces or volumes, depending on the nature of the broken artefacts. Presented here is a fast, unified method that combines curve matching techniques with a surface matching algorithm to estimate the positioning and respective matching error for the joining of three-dimensional fragmented objects. Combining both aspects of fragment matching, essentially eliminates most of the ambiguities present in each one of the matching problem categories and helps provide more accurate results with low computational cost.
Downloads: the paper
Reference: BibTex
-
Reconstruction of Three-dimensional Objects through Matching of their Parts
G. Papaioannou, E.A. Karabassi, T. Theoharis, IEEE Transactions on Pattern Analysis and Machine Intelligence, 24(1), pp.114-124, 2002.
Abstract. The problem of reassembling an object from its parts or fragments has never been addressed with a unified computational approach, which depends on the pure geometric form of the parts and not application-specific features. We propose a method for the automatic reconstruction of a model based on the geometry of its parts, which may be computer-generated models or range-scanned models. The matching process can benefit from any other external constraint imposed by the specific application.
Downloads: the paper
Reference: BibTex
-
Virtual Archaeologist: Assembling the Past
G. Papaioannou, E.A. Karabassi, T. Theoharis, IEEE Computer Graphics and Applications, 21(2), pp. 53-59, 2001.
Abstract. This article describes a semi-automatic system for the reconstruction of archaeological finds from their fragments. Virtual Archaeologist is a system that uses computer graphics to calculate a measure of complementary matching between scanned data and employs optimization algorithms in order to estimate the correct relative pose between fragments and cluster those fragments that belong to the same entity.
Downloads: the paper
Reference: BibTex
-
Segmentation and Surface Characterization of Arbitrary 3D Meshes for Object Reconstruction and Recognition
G. Papaioannou, E.A. Karabassi, T. Theoharis, Proc. IEEE International Conference on Pattern Recognition, 2000, pp. 734-737.
Abstract. Polygonal models are the most common representation of structured 3D data in computer graphics, pattern recognition and machine vision. The method presented here automatically identifies and labels all compact surface regions of a polygonal mesh, visible or not, and extracts valuable invariant features regarding their geometric attributes. A method that is independent of the mesh topology is also presented for the surface bumpiness estimation and the identification of coarse surface regions.
Downloads: the paper
Reference: BibTex
-
Automatic Reconstruction of Archaeological Finds – A Graphics Approach
G. Papaioannou, E.A. Karabassi, T. Theoharis, Proc. 4th International Conference on Computer Graphics and Artificial Intelligence (3IA' 2000), p.p. 117-125, 2000.
Abstract. Reconstruction of archaeological finds from fragments, is a tedious task requiring many hours of work from the archaeologists and restoration personnel. Up to now, computers have significantly simplified this work by providing tools for the data encoding, storage, classification and visualisation in some cases. In this paper we go one step further by presenting a semi-automatic procedure for the full reconstruction of the original objects using computer graphics and artificial intelligence algorithms on geometrical information.
Downloads: the paper
Reference: BibTex
-
Automatic Reconstruction of Objects from their Parts and its Application to Digital Archaeology
G. Papaioannou, Ph.D. Dissertation (in Greek), Department of Informatics and Telecommunications, National and Kapodestrian University of Athens, Supervisor: T. Theoharis, 2001.
Abstract. In this PhD thesis, a general computational method for the automatic reconstruction and assemblage of three-dimensional objects from parts or fragments was proposed and developed. This reconstruction problem may be regarded as a generalization of the jigsaw puzzle, where the number and shape of the pieces are unknown and some parts may be missing or damaged. The above problem arises mostly in archaeological reconstruction and restoration nut also in other scientific fields. The proposed method operates on the of digitized part models, estimates a measure of the complementary fitting between different pieces and forms clusters of these (properly aligned) models that represent the final reconstructed entities. The method relies on the geometry of the digitized data to solve the problem. However, additional available information, such as material attributes or structural details or patterns can improve both the method's performance and the quality of the results. The matching and data analysis algorithms used take advantage of graphics hardware to achieve high performance and combine techniques from the computer graphics, pattern analysis and optimization theory fields.
Downloads: Ph.D. dissertation (in Greek)