Infrared small target detection based on the self-information map

The achievement of high detection probability and low false alarm probability is a bottleneck problem in the detection of infrared small targets. Thus, a new small target detection approach which integrates the concept of self-information map (SINM) with the adaptive thresholding method followed by a region growing technique is proposed in this paper. The concepts of local signal-to-noise ratio, region nonuniformity, detection probability, and false alarm probability are used to evaluate the performance. The experiment results obtained from qualitative and quantitative comparisons testify to the robustness of the approach presented under different conditions.

Research highlights
► A novel infrared small targets detection approach is on the basis of the concept of self-information map.
► The concept of SINM originates from the visual mechanism of flying insects, and SINM is constructed by a Parzen window function.
► The adaptive thresholding method followed by a region growing technique is used to segment small targets.
► The new algorithm yields high detection probability and low false alarm probability.