【作者】 刘冬梅；

【导师】 赵宇明；

【作者基本信息】 上海交通大学 ， 模式识别与智能系统， 2009， 硕士

【摘要】 动态范围图像的显示是图像处理、机器视觉、模式识别等学科都十分关注的问题,属于一个交叉性的问题,对它的研究有着重要的和广泛的意义。尤其是随着数码技术的不断发展,不论是静止图像还是视频都对图像的显示提出了较高的要求。在这个问题中,高动态范围图像的显示是项重要的技术。本文就这项技术,做了广泛的调查研究,查阅了国内外相关的资料信息,总结了现有的几种具有代表性的色阶重建算法,并通过在梯度域上对图像动态范围进行压缩调整,使得高动态范围的图像能够在常规显示设备上得以良好地显示,使图像在观察时与真实场景一致。该算法能够解决的是静态图像的色阶重建问题。算法根据图像的梯度信息能够自适应地得到较好的效果。该算法基于梯度高斯金字塔,依据梯度的大小进行自适应的梯度压缩,然后通过快速傅立叶变换求解泊松方程计算图像的亮度图像。实验结果表明,本文算法能够较好地避免了光晕,同时保持了图像的细节。算法的流程简单,处理速度快,适用性强,不仅能适用于PC电脑处理高动态范围图像,经过改进之后,还能适用于嵌入式设备中的色阶重建过程。更多还原

【Abstract】 Today displaying a range image is becoming more and more important in many fields, such as digital photography,film processing,scientific image enhancement,virtual reality,computer vision,pattern recognition and so on. With the development of the digital technology, there is a higher requirement for the image displaying technology in both digital image filed and video filed. How to get a better visual output in a low dynamic range output device is now an important subject.According to the subject, this paper presents a summary of previous results and proposes a local adaptation algorithm of compressing high dynamic range images based on gradient attenuation to fit conventional display devices that are only capable of outputting a low dynamic range.This algorithm can solve the digital image tone mapping problem and have excellent displaying result adaptively according to the gradient. Based on Gaussian pyramid, the algorithm compresses the gradient according to the amplitudes, and then gets the luminance image through FFT to solve the Poisson function. Experimental results show the algorithm of manipulating the gradient field of luminance image by attenuating the large gradients’magnitudes can preserve fine details avoiding common artifacts, such as halos, gradient reversals, or loss of local contrast. In addition, this algorithm can be not only used in personal computer but also in embedded equipments.更多还原