【作者】 罗涛；

【导师】 宋刚；

【作者基本信息】 山东大学 ， 电路与系统， 2008， 硕士

【摘要】 随着计算机性能的不断提高、多媒体技术不断完善,特别是彩色成像设备的不断改进,彩色图像的应用越来越广泛,基于彩色图像的处理技术也显得日益重要。在显示器领域,液晶显示器虽然有着轻薄、省电、无辐射、高清高亮、接口丰富等优点,但在色彩表现方面存在先天的缺陷,因此液晶显示器彩色图像的增强技术得到越来越多人的关注。针对液晶显示器色彩表现差的问题,我们采用对LCD图像处理单元进行改进的方法,主要是对彩色图像的亮度、对比度及饱和度进行处理,以达到增强LCD色彩表现力的目的。在介绍课题背景及意义的基础上,对彩色图像增强处理的发展动态进行了简要说明,并介绍了人类视觉基本原理、人眼视觉特性以及色度学的基础知识,为后续工作打下了理论基础。传统的灰度图像空域增强算法包括线性变换、非线性变换、直方图修正、平滑滤波、锐化滤波等。将这些算法应用到彩色图像,并分析了这些算法的优缺点。对彩色图像的亮度分量,提出了基于小波变换的彩色图像增强方法。在符合人眼视觉特性的HSV空间中,对亮度分量进行小波分解后,根据低频系数和高频系数的不同特性,对它们分别进行处理。对表示图像细节的高频系数,根据局部方差的大小,将之分为低对比度区(平滑区)、中等对比度区和高对比度区。根据人眼对图像平滑区域噪声比较敏感这一视觉特性,对平滑区不进行处理;对中等对比度区域和高对比度区域采用不同增强系数的非线性反锐化掩模方法进行处理,这样既可以增强图像细节,又能够抑制噪声。对表述图像概貌的低频系数进行非线性函数映射,以增强图像整体对比度和亮度。对彩色图像的饱和度分量进行简单的指数拉伸。处理后彩色图像不管是从亮度方面、细节方面,还是色彩表现方面都有了显著的增强,得到了较好的视觉效果。同时,根据饱和度分割算法,提出了一种新的与亮度关联的彩色图像饱和度增强算法。在HSV色彩空间,利用色度空间分割算法,将饱和度分量分割为饱和度很高、饱和度高、饱和度低和饱和度很低四个区域。按照人眼对自然界不同饱和度颜色的感应,对不同的饱和度区域利用调整曲线进行处理。对饱和度高、饱和度低两个区域采用正增强方法提高饱和度;对饱和度很高区域采用负增强方法,适当降低饱和度。然后,根据人眼视觉特性,利用亮度-饱和度相关系数引入亮度反馈,对饱和度进行进一步处理,使之更符合人眼感知特性。通过选择适当的参数,可以使饱和度失真的图像得到较好的增强。更多还原

【Abstract】 With the development of the Computer capability, the maturity of multimedia technology, and especially the development of the color imaging device, color image processing technology becomes more and more important. Especially in display field, the newly emerging LCD display has many advantages such as portability, less consumption of power, no radiation, high definition, high brightness and abundant interfaces. But due to the defect of the color expression, more and more people devoted to the color image enhancement technology of LCD. According to this problem, this paper mainly improves the LCD image processing cell by processing the luminance, contrast and saturation of color image to improve the color expressive ability.Based on the introduction of the background and significance of the subject, this paper explaines the development trend of color image enhancement briefly. Then introduced the basic principles of human vision, the factor of human vision and basic knowledge of colorimetry.The conventional spatial domain enhancement methods of the gray image include linear transformation, nonlinear transformation, histogram correction, smoothing filtering and sharp filtering. In this paper these algorithms are applied to color image processing and their advantages and disadvantages are analyzed.For luminance component, a new method of color image enhancement based on the wavelet transform is proposed. In HSV space which accord with human visual system, the low frequency coefficient and high frequency coefficient of the luminance component which are decomposed by wavelet are processed respectively according to their different properties. For the high frequency coefficient which expresses image details, we use the local variance to divide the high frequency component into smooth region, medium contrast region and high contrast region. Based on the human visual properties, in the smooth region which is sensitive to noise, we do nothing. In medium contrast region and high contrast region, we use the nonlinear unsharp masking method which has different enhance coefficient, this method can enhance the local information of images and also can restrain the noise. For the low frequency coefficient which expresse the general appearance of image, we use the nonlinear mapping function to enhance the global contrast and luminance of image. The saturation component of color image is enhanced by an exponential function. After processing, the image’s luminance, details and color expression are improved.In addition, according to saturation segmentation algorithm, a new saturation enhancement algorithm with luminance component is proposed. In HSV space, the saturation component is divided into four regions: very-high saturation, high saturation, low saturation and very-low saturation. The different saturation regions are processed using a different adjustment curve because the different visual induction for saturation colour. For the high saturation and low saturation regions, we use positive enhancement to improve the saturation, and for the very-high saturation region, we use negative enhancement to properly reduce the saturation. Finally, from the luminance-saturation correlation coefficient, the luminance feedback were proposed for a further treatment of the saturation component, thus the saturation is more accord with the human visual perception. Then the color saturation distortion image can be enhanced by choosing proper parameters.更多还原