Fast local Laplacian filter-based tone mapping operator via brightness, contrast, and color enhancements using S-shape functions
摘要
A high-dynamic-range (HDR) image accommodates a wide range of brightness information of a scene. A tone-mapping operator (TMO) compresses the dynamic range of an HDR image to display the HDR image on a low-dynamic-range (LDR)-output device. Paris et al. proposed a TMO that uses a local Laplacian filter (LLF). LLF manipulates the Laplacian pyramid coefficients of the input image for edge-aware image processing. Paris et al.’s TMO outputs a comparably high-quality and natural-looking image. In addition, it can be accelerated by using a fast local Laplacian filter (FLLF), which is a fast approximation version of LLF. However, it suffers from loss of details in bright regions. In this paper, we propose a new TMO using FLLF and S-shape functions. In it, the dynamic range compression using FLLF is applied to the input HDR image. Then, to obtain an image with natural brightness, the brightness is adjusted using S-shape functions. After that, local contrast and color enhancements using S-shape functions are applied to the brightness-adjusted image. The contrast and color-enhanced images are fused using the equi-hue plane so that the local contrast and color enhancement effects are properly reflected on the final output image. The proposed TMO can generate natural-looking and high-quality tone-mapped images with suppressed loss of details. Through qualitative and quantitative evaluations, we demonstrate the effectiveness of our proposed TMO.