GLoG: Laplacian of Gaussian for Spatial Pattern Detection in Spatio-Temporal Data

Area of Science:

• Computer Vision
• Graph Signal Processing
• Data Analysis

Background:

• Boundary detection is crucial for analyzing static and dynamic data in image processing and computer vision.
• Existing methods often struggle with complex spatio-temporal datasets.

Purpose of the Study:

• To propose a novel boundary detection filter for graphs, extending image processing techniques to spatio-temporal data.
• To enable the analysis of spatial patterns and temporal randomness in graph-structured data.

Main Methods:

• Developed a graph-based filter analogous to the Laplacian of Gaussian (LoG) edge detector.
• Applied Graph Signal Processing (GSP) theory for filter construction.
• Introduced entropy calculation for time slices to measure randomness.

Main Results:

• The proposed filter effectively reveals significant spatial patterns within graph data.
• Entropy of time slices successfully identifies expected and unexpected temporal phenomena.
• Demonstrated effectiveness on both synthetic and real-world spatio-temporal datasets.

Conclusions:

• The novel graph boundary detection filter enhances visual analytics for spatio-temporal data.
• The method provides a robust mechanism for uncovering complex spatial and temporal insights.
• This approach offers a valuable tool for researchers and practitioners in data analysis.