Gradient-Constraint Super-Resolution Reconstruction Method Serving for Infrared Target Detection

Abstract

Infrared images generally suffer from insufficient resolution, blurred edge, and low contrast, which increases the difficulty of infrared target detection. To solve this problem, a multiframe image super-resolution (SR) algorithm based on edge gradient regularization is proposed. In order to improve the resolution of the infrared image while simultaneously improving the visual saliency of weak infrared target, an edge-preserving regularization term is designed and introduced into the solving process of Bayesian problem. On this basis, an image enhancement method based on gradient-constraint decomposition is proposed to improve the contrast of the infrared image, which further lays a foundation for infrared target detection. To verify the effectiveness of the algorithm, experiments are respectively carried out on FLIR infrared datasets, real infrared images captured statically and real infrared images captured by a UAV equipped with an infrared camera. The experimental results demonstrate that the proposed method is capable of generating high-resolution images with good performance in terms of edge preservation and detail enhancement. Compared with the original input infrared images, the edge saliency of the targets inside the SR reconstructed infrared images is significantly enhanced, and thus provides a very promising image preprocessing method for infrared target detection.