Multi-field Pattern Matching based on Sparse Feature Sampling

Abstract

We present an approach to pattern matching in 3D multi-field scalar data. Existing pattern matching algorithms work on single scalar or vector fields only, yet many numerical simulations output multi-field data where only a joint analysis of multiple fields describes the underlying phenomenon fully. Our method takes this into account by bundling information from multiple fields into the description of a pattern. First, we extract a sparse set of features for each 3D scalar field using the 3D SIFT algorithm (Scale-Invariant Feature Transform). This allows for a memory-saving description of prominent features in the data with invariance to translation, rotation, and scaling. Second, the user defines a pattern as a set of SIFT features in multiple fields by e.g. brushing a region of interest. Third, we locate and rank matching patterns in the entire data set. Experiments show that our algorithm is efficient in terms of required memory and computational efforts.

IsoExplorer: an isosurface-driven framework for 3D shape analysis of biomedical volume data

“…Bruckner and Möller (2010) quantified the similarity between isosurfaces from an information-theoretic perspective by mutual information. Furthermore, Wang et al (2015) introduced 3D SIFT to facilitate user-defined pattern matching in 3D multi-field scalar data.…”

“…In these three datasets, the averaged Top-5 accuracy of IsoExplorer and FlowNet are 84% and 57%, respectively. Compared to traditional knowledge-driven 3D descriptors, which focus on specific tasks (e.g., contour tree for symmetry analysis Thomas and Natarajan 2011, 3D SIFT for pattern matching Wang et al 2015), data-driven 3D descriptors are more generic for different tasks and require no or little knowledge about data (Rostami et al 2019). During retrieval between different volume data, users can establish connections between shapes scanned at different resolutions and search the shape database for target shapes as keywords to find similar shapes.…”

Multivariate spatial data visualization: a survey

“…Interactive classification [65] [67], [19], [18], [40], [39], [68], [41] Data mining [26], [66], [50], [63], [60], [62] Topological structures [43], [12], [6], [16], [7], [64], [54] Fusion visualization Data fusion [53], [19], [32], [28], [7], [24], [41], [13] Feature fusion [38], [18], [11] Image fusion [4], [66], [49] Correlation analysis Voxels [44], [63] Variables [58], [3], [13] Numerical values [22], [27], [6], [39] Features [47], [59] Value-variable [3], [39] 2 Feature Classification…”

Global Feature Tracking and Similarity Estimation in Time‐Dependent Scalar Fields

“…The basic premise of pattern matching is to find regions or features that are similar to a designed pattern or a selected region/feature. Such methods exist for a large variety of data types such as images [Low04], geometry [MPWC13], scalar fields [KWKS11, SSW14, SSW15, TN11, TN13, TN14], vector fields [ES03, HEWK03, BHSH14], and multi‐fields [WSW16]. All of these methods address single time steps only, and are not adequate for finding spatio‐temporal similarities: given a pattern, one may find similar features in a number of time steps, but this neglects any temporal evolution, since a progressively changing feature matches a pattern only for a certain amount of time.…”