Research Of High Dynamic Range Image Generation Based On Inverse Tone Mapping

In recent years,there has been a growing demand for high-quality display devices,and high dynamic range technologies have received more attention from researchers.A lot of manufactures have released electronic products which support high dynamic range technologies to satisfy the need of consumers.High dynamic range images can restore lighting conditions in nature scene more realistically,and reveal vivid visual effects.This thesis mainly based on inverse tone mapping,explores the methods of image preprocessing and dynamic range expansion,which also elaborate and implements simulations on proposed high dynamic range image generation algorithms.Taking into account the non-linearity between the real scene captured by the camera and light conditions of the stored image,in order to restore lighting conditions closest to the real scene,this thesis uses inverse gamma function,which restore lighting conditions of stored images to lighting conditions captured by the camera.Bilateral filter is used for conventional image denoising methods,but it cannot smooth images properly under the condition of small parameter values.This thesis introduces a weight least squared filter to remove noise from the image.Due to the restriction of light conditions of natural scene,camera capturing may produce highlight or saturated areas,which needs to be preprocessed.Firstly,this thesis detects highlight areas based on the low pass filter.For a large number of saturated areas of some images,this thesis used a new method based on threshold to detect these areas.Traditional bilateral filter cannot handle highlight areas properly and it is very slow,so this thesis uses bilateral grid to preprocess highlight areas of the image.The grayscale image and highlight areas are used as the original data,and the preprocessing image is generated using the bilateral grid.Then it mixes the input image and the preprocessing image to generate a preprocessed image.Finally,based on the method of inverse tone mapping,this thesis makes a dynamic range expansion of the high-quality preprocessed image.In this thesis,the inverse tone mapping is realized by human visual system or piecewise linear function.First of all,this thesis uses human visual system to expand the dynamic range of the preprocessed image.In order to simplify the dynamic range expansion method,this thesis proposes a light related linear function to expand the dynamic range of the image.Based on above research,this thesis innovatively proposes a new piecewise linear function to expand the dynamic range of the image and generate the high dynamic range image finally.Experiments use a lot of methods to assess the results and efficiency of this thesis.Both subjective assessment and objective assessment confirm that the algorithm of this thesis can produce less distortion,keep color fidelity and reveal more details.This algorithm has high processing efficiency,and can obtain expanded images with larger dynamic range and generate high-quality high dynamic range images.