Infrared target detection based on image entropy

Abstract

In the infrared target detection, the size of structural element tend to be set as pre-determined size, when using the traditional morphological filter based on Top-Hat transform for suppressing the background of the digital infrared image. However, if the target size is unknown or keep changing in the successive frames, the fixed size of the structural element can not meet the detection requirement. To address this issue, we use both Top-Hat Transform and Ostu threshold segmentation algorithm. When the size of structural element is changing continuously, we obtain the different detection results. By analyzing the entropies of the different detection images, a new infrared target detection method was adopted based on the entropy of the detection image. The experimental results prove that this approach can find the best size of structural element, and well detect the target.