Modeling possible defocus and vibration distortions of infra-red images

Infra-red (IR) sensors are very common devices for qualitative night observation and intrusion detection. In many applications the IR sensor is being operated from a flying and vibrating platform. In addition, in many applications the sensor is required to detect close as well as remote objects. Placing the sensor over a vibrating platform and scanning close and remote objects often creates corresponding distortions over the captured IR image. A general statistical model for describing the obtained image after incorporating the above-mentioned distortions, is presented. The observed terrain itself is statistically modeled by the first order 2D Markov process known from the literature. The resulting statistical model takes into account various parameters such as the range between the camera and the terrain, the correlation length of the terrain, the type of the vibration and its specifications and the amount of defocus. Special numerical techniques may be applied in order to convert the statistical model into a real 2-D IR image. The formulation derived by this paper may be incorporated in IR simulators that tend to examine the performance of detection algorithms in front of reliable IR images, which take into account various feasible distortions.