A Fast and Robust Automatic Object Detection Algorithm to Detect Small Objects in Infrared Images

Automatic object detection in infrared images is a vital task for many military defense systems. The high detection rate and low false detection rate of this phase directly affect the performance of the following algorithms in the system as well as the general performance of the system. In this work, a fast and robust algorithm is proposed for detection of small and high intensity objects in infrared scenes. Top-hat transformation and mean filter was used to increase the visibility of the objects, and a two-layer thresholding algorithm was introduced to calculate the object sizes more accurately. Finally, small objects extracted by using post processing methods.