Moving persons tracking from infrared image sequences

A. Andonova1, A. Bekiarski1 and S. Pleshkova-Bekiarska1

1Technical University of Sofia, Bulgaria

Keywords: thermal radiation, thermohraphy, images processing
property: TUS
material: paper

The development of methods and algorithms for moving person detecting and tracking is very popular for visual images processing systems. In the range of infrared images, capturing and visualization from thermo visual sensors or thermo cameras of objects and persons can be combined with the effective methods and algorithms for moving person detecting and tracking. There are many methods and tools for processing the thermo vision images, which are based on the general principles for processing visual images, taking into account the specific features of thermo vision images, particularly the features of recording thermal radiation and its visualization in the form of pseudo visible image. These general principles of existing methods of processing thermo vision images can be used for developing of new methods orientated in specific direction of processing of thermo vision images - detecting and tracking objects and people for the purpose of tracking information systems for customs control and counter-terrorism. The goal of this article is to use thermo vision sensors as the main component of thermo vision cameras that form the thermo vision images by registering the thermal radiation in the infrared spectrum instead of the visible radiation from object and the environment. The infrared image sequences, taken from thermo visual camera, are processed with the proposed algorithm first to detect the existence of one or more persons in the infrared image frames and then to track the moving persons. The proposed algorithm is presented with appropriate equations describing the basic principles of infrared images preprocessing and human body separation. Then the proposed tracking method is described to show its possibility of finding the place of tracking person in the sequence of infrared thermal images frames. The proposed algorithm is tested with the appropriate simulations and real infrared images of moving persons. The results are presented in form of infrared test images, graphical representations of tracking trajectories and tables shown the precision of finding and tracking the moving persons.

  1. A. Apatean, A. Rogozan, A. Bensrhair, Obstacle Recognition using Multiple Kernel in Visible and Infrared Images, IEEE Intelligent Vehicles Symposium, 370-375 (2009)

  2. R. Hammoud, (Ed.), Augmented Vision Perception in Infrared Algo-rithms and Applied Systems, (Springer 2009), pp.371-401

     Official Sponsors

uni Anter_logo Tziolas_logo_ linseis netzsch