Development of different approaches for object tracking and classification in thermal videos

Abstract

In todayand#8223;s modern technology the thermal camera plays a major role in all security applications, surveillance applications etc. and also in pandemic situations the thermal camera plays a major role, because of its nature of picturizing the scene based on the temperature emitted by the objects in a scene. Hence regardless of any weather conditions and lighting conditions, an object can be visualized. But still thermal images have constraints like no texture or colour information, a greater number of dead pixels, low resolution, and only noticeable visual colour patterns in case of any temperature variations. So, in general, there is huge demand for effective object tracking and classification system in thermal videos. The proposed methods are TFM (Tri Feature Matrix) based object tracking, STP (Spatial Temperature Pattern) and TTP (Temporal Temperature Pattern) based object classification and FOT (Forward Oriented Temperature) and BOT (Backward Oriented Temperature) based object classification. Object tracking in thermal videos is a challenging task to track object without identity mismatch because of lack of clear edges in thermal frames and tracking in case of occlusion is another challenge. The TFM based object tracker uses TFM as an object descriptor which is used to uniquely identify and represent objects in thermal images. TFM is an aggregation of three features such as distance between convex hull points, distance between convex deficiency points and Fourier Descriptor. TFM is represented in more compact way as a triple matrix. It is an accurate descriptor suitable for tracking objects in thermal video sequences without an identity switch newline