Research On CRF Estimation And HDR Imaging Generation Method Based On Single Image

High dynamic range(HDR)imaging is the term given to the capture,storage,manipulation,transmission,and display of images that more accurately represent the wide range of real-world lighting levels.From the experience of more than 20 years of static HDR imagery,HDR is finally ready to enter the “mainstream” of imaging technology.Researchers have proposed a number of methods to acquire HDR image.The HDR imaging methods can be divided into single exposure and multi-exposure.The multi-exposure HDR imaging uses several frames with different exposure time to fuse the HDR image,and we can generate HDR image only using a single input image.Most single-image based HDR image generation methods expand dynamic range only from bright areas which reduces details visible in shaded areas and induces artifacts at the edge of bright areas.This inspires us to exploit an approach expanding dynamic range both from highlight region and shaded region.In this paper,we propose a method for generating high dynamic range(HDR)radiance maps from a single low dynamic range(LDR)image based on its camera response function(CRF).The proposed method achieves this goal by performing ICRF on image intensity to recover the image irradiance which is taken as HDR image.The algorithm is based on the estimated inverse camera response function(ICRF)and the locally weighted linear regression(LWLR)used in conducting the ICRF on the single LDR image.The main contributions of the proposed method consist in two aspects: the Parzen window function,mixture Gaussian model or triangular nuclear density estimation used in the estimation of ICRF and the utilization of locally weighted linear regression(LWLR)with ICRF in HDR image generation.The experimental section of this paper compares many HDR imaging methods and evaluates them separately from subjective and objective image quality.Extensive experimental results show that the approach induces less contrast distortion and produces high visual quality HDR image.