| The wide range of light intensities and rich color could be represented in a high dynamic range (HDR) image in real-life scenarios under a variety of lighting conditions. However, facing high dynamic range scenes, the usual cameras are encountered by the limitations of dynamic range and are unable to obtain a HDR image. Meanwhile, the standard display cannot reproduce a HDR image directly. In recent years, domestic and foreign scholars have made in-depth research on synthesis and visualization for HDR images, and a lot of methods have been proposed. However, most of the algorithms improve effectiveness at the expense of efficiency, and cannot be directly applied to the intelligent mobile terminals which have a higher and higher requirement for real-time image processing. In this dissertation, the study is focused on the key technologies of fast synthesis and visualization method for high dynamic range images, including the rapid calibration method for camera response function, the fast and effective method to construct radiance maps and the high efficiency tone mapping operators based on guided filter. The main work and contributions of this dissertation are as follows:(1) A rapid calibration method for camera response function has been presented to solve the low-efficiency of traditional calibration algorithms in synthesis of high dynamic range images. Firstly, according to the relation among multi-exposure images, a new least square error function is proposed to reduce the dimensions of calibration equation. Then robust QR decomposition is substituted for singular value decomposition algorithm to accelerate the calculation of camera response function. Compared with the traditional methods, the proposed method is with lower computational complexity, higher speed and higher precision as well.(2) An effective method has been presented to generate high dynamic range images based on multi-exposure techniques. On the basis of the above rapid calibration method, a kind of effective automated sample selection strategy is designed to minimize the effect of optical blur of the imaging system. Other than triangle weight function, Gaussian weight function is used to eliminate ambient noise and maintain the efficiency for constructing a radiance map that covers the full dynamic range by combining multi-exposure images of the same scene.(3) Through analyzing and proving the low efficiency and color cast of iCAM06 algorithm, an algorithm of tone mapping based on guided filter and hierarchical model has been proposed for HDR images. A HDR image is firstly decomposed into a base layer and a detail layer by an improved guided filtering method. For the base layer, its dynamic range is compressed by the simulated camera response function. For the detail layer, it is enhanced to produce more fine structures by using the guided image filter. The color balance correction method is adopted to reduce color cast. The experiments on real HDR images demonstrate that the proposed algorithm has better brightness, contrast, and visibility without blurring or undesired artifacts.(4) A novel tone mapping algorithm based on principal component analysis (PCA) and image enhancement has been proposed to balance effectiveness and efficiency in the previous algorithms. Firstly, the adaptive gamma correction is adopted to compress the dynamic range in dark and light areas according to the average brightness of HDR images. Then R, G, B color channels of the image are converted to the orthogonal space by PCA method. The dynamic range of the first principal component is compressed by the curve to simulate camera response function (CRF). Combining with two weighted chrominance components, the result is transformed to a new image by PCA inverse transform. Finally, a fast guided filtering method for image enhancement is implemented to improve image contrast while avoiding halo phenomenon. |
PDF Full Text Request