Variational Template Matching Method Targets Anomaly Detection in Small-Data Settings

A new arXiv preprint proposes combining classical template matching with variational techniques and statistical fusion to detect anomalies in patterned images. The authors argue that deep learning is often too costly or impractical when training data is scarce, while traditional template matching is interpretable but brittle to changes in scale and geometry. The method aims to keep the simplicity of template-based approaches while improving robustness to such variations.