Vis Comput DOI 10.1007/s00371-017-1395-4 ORIGINAL ARTICLE Adaptive compression of animated meshes by exploiting orthogonal iterations Aris S. Lalos 1 · Andreas A. Vasilakis 2 · Anastasios Dimas 1 · Konstantinos Moustakas 1 © Springer-Verlag Berlin Heidelberg 2017 Abstract We introduce a novel approach to support fast and efficient lossy compression of arbitrary animation sequences ideally suited for real-time scenarios, such as streaming and content creation applications, where input is not known a priori and is dynamically generated. The presented method exploits temporal coherence by altering the principal component analysis (PCA) procedure from a batch- to an adaptive-basis aiming to simultaneously sup- port three important objectives: fast compression times, reduced memory requirements and high-quality reproduc- tion results. A dynamic compression pipeline is presented that can efficiently approximate the k -largest PCA bases based on the previous iteration (frame block) at a signifi- cantly lower complexity than directly computing the singular value decomposition. To avoid errors when a fixed number of basis vectors are used for all frame blocks, a flexible solution that automatically identifies the optimal subspace size for each one is also offered. An extensive experimental study is finally offered, showing that the proposed methods are supe- rior in terms of performance as compared to several direct PCA-based schemes while, at the same time, achieves plau- sible reconstruction output despite the constraints posed by arbitrarily complex animated scenarios. Electronic supplementary material The online version of this article (doi:10.1007/s00371-017-1395-4) contains supplementary material, which is available to authorized users. B Aris S. Lalos aris.lalos@ece.upatras.gr 1 Electrical and Computer Engineering Department, University of Patras, Patras, Greece 2 Information Technologies Institute, Centre for Research and Technology Hellas, Athens, Greece Keywords 3D animated meshes · Orthogonal iterations · Online compression 1 Introduction With the rapid advances in high-performance computing, scanning operations and content creation tools, the out- put data are expanding rapidly generating massive datasets. While this can be mitigated by applying compression tech- niques to the data being archived or transferred, the tremen- dous computing resources required bring tough challenges to be solved. Recently, there has been increasing interest on acquiring, processing, storing and transmitting 3D ani- mated meshes facilitating several real-time applications (e.g., Microsoft Holoportation, an immersive telepresence sys- tem). Throughout the years, numerous approaches have been proposed improving more or less some of the key anima- tion compression characteristics [16]: encoding–decoding requirements, reconstruction quality and compression rates. Without loss of generality, these methods can be classified either as local - or global -based, depending on the frame win- dow analysis taken for compressing the animation sequence. The main benefit of the global approaches is an improved compression rate by analyzing the overall motion coherence, whereas the local ones focus on local frame-to-frame transi- tions allowing low-latency streaming. Skinning as well as principal component analysis (PCA) can be considered as the most well-known global methods for providing efficient compact representations of rigid and highly deformable animations, respectively. Though a large variety of different strategies have been introduced in both skinning [8] and PCA [16], all of them suffer from exces- sive computational requirements, strongly dependent on the 123