Infrared Dim Small Target Detection of Ship on the Sea Based on Improved RPCA

In the infrared imaging of the sea and river surface, ships are often very small and difficult to identify, and lack of many effective information makes it difficult to identify ships. To address the aforementioned challenges, a weak and small ship detection method based on RPCA combined directional derivative penalty is proposed. First of all, to distinguish the ship from the river in the infrared image, we use the difference of the distribution of the first-order directional derivative in the original image to build a prominent image. Then, the weight matrix corresponding to saliency mapping is combined with
$\boldsymbol{l}_{\mathbf{1}}$
norm to strengthen the constraint on sparse targets. Finally, the alternative direction multiplier method(ADMM) is used to solve the model. It can well detect the position of the ship in the infrared image of the sea.