The Framework of Infrared Video Mining Based on Topic Model
Lin Liu, Lin Tang, Hong Li, Shaowen Yao · 2014
We proposed a framework of infrared video mining based on topic model. It aims to learn motion patterns for a crowded and complicated infrared scene. After video preprocessing, motion features are extracted from each pair of consecutive frames at first, and quantized into visual words. Motion pattern are modeled as distributions over visual words in topic model. Experiments about BOVW demonstrate the feasibility of the framework.