Research Of High Dynamic Range Image Enhancement Based On Visual Adaptation Mechanism

With our increasing demand of the Image quality, High Dynamic Range Image(High Dynamic Range Image, HDRI) is being widely used due to its richer cover scene information and lower cost. However, the dynamic ranges of most current output devices are too restricted to display the whole information HDR images contain. As a result, it is necessary to compress the dynamic ranges of HDR images as much as possible without losing the original information in order to match traditional displayers. It is known that human visual system is able to perceive a hug range of intensities, from the midday sun to the stars at night. Based on human visual physiology, retinal light and dark adaptation system and existing psychophysics experiments, this dissertation proposes two algorithms to compress the dynamic range of HDR images, both of which are static, local and almost automatic.The first model is based on an existing framework proposed by Tumblin, in which a forward model and a reverse model could be used to compute the output image in that input image(original scene) and output image(display image)have perceptual match. In our model, the mechanism of photosensitive pigment bleaching is used to compute the response of photoreceptors and a chromatic adaptation model is added up to Tumblin's framework.We present an innovative arithmetic framework to compress the range of HDR image in our second model. In this model we pay more attention to simulating the physiological process of eyes adaptation. And the major part of our model can be divided into three components, which include photoreceptors luminous model, photoreceptors response model and ganglion cells function. Besides, synthesis of rods and cones response signals as well as color stable are also taken into considerable consideration.