High dynamic range infrared image enhancement via central linearity suppression guided filter and local high radiation pass filter

Abstract

Existing high dynamic range (HDR) infrared image enhancement methods have limitations of local over-exposure, edge artifacts or target missing. In this paper, we present an effective HDR infrared image enhancement method to addresses these issues simultaneously, which can achieve visually friendly results in many scenes. Specifically, we first introduce an inverted contrast enhancement module, which solve over-exposure by stretching the dynamic range of high radiation regions as evenly as possible. Next, we adjust the contrast with gamma function and use the result as basic image. Then, we propose a central linearity suppression guided filter (CLSGF) by suppressing the linear fitting level of the central pixel of the window in the guided filter. The CLSGF is used as detail layering filter to effectively suppress edge artifacts in detail enhancement stage. Subsequently, we observe the visual relationship between the detail layer and the inverse detail layer, and design a local high radiation pass filter (LHRPF) using Gaussian function. It senses the local radiation variations of image and is used to highlight the local high radiation details/targets. Finally, a linear combination of the basic image, the detail layer and the LHR image yields enhanced result, which fully expose the scene contents and provide high contrast and stable visual effects. Extensive experiments demonstrate the effectiveness of the proposed method.