Long range automatic detection of small targets in sequences of noisy thermal infrared images

In this paper, an approach to the automatic detection of vehicles at long range using sequences of thermal infrared images is presented. The vehicles in the sequences can be either moving or stationary. The sensor can also be mounted on a moving platform. The targetarea in the images is very small, typically less than 10 pixels on target. The proposed method consists of two independent parts. The first part seeks for possible targets in individual images and then merges the results for a subsequence of images. The decision for this part of the algorithm is based on temporal and spatial consistency of the targets through the considered image subsequence. The second part of the algorithm specifically focuses on finding moving objects in the scene. Clearly, as the sensor may itself be moving too, the effect of this motion on the images has to be eliminated first. This was done using a model based registration technique. The algorithm proposed in this paper was implemented and tested on a set of 7 image sequences obtained from different sensors under diverse operational circumstances. The images in these sequences are mostly highly cluttered and noisy. Results show the two parts of the algorithm to be quite complementary. For some sequences, the first part yields good results (high ratio of detection probability to false alarm rate) while, for other sequences, the second part gives the best results.