Vol. 47, Issue 3, pp. 383-394

Abstract

A new method for passive ranging using image size measurements from one sensor has been presented. The method relies on pixel filtration with histogram-based thresholding, followed by intensity and gradient magnitude analysis. Its efficiency and robustness were assessed in real infrared surveillance sequences, and it has proved to lead to better results than non-filtering techniques. The object distance estimation mean relative error does not exceed 3%, which implies that the suggested method enables precise range estimation based on object size measurements. To maximize the benefits of the suggested method, Kalman filter has been included in the algorithm in order to overcome fluctuations of the estimated object size.