【作者】 陈强；

【导师】 陈贺新；

【作者基本信息】 吉林大学 ， 通信与信息系统， 2006， 博士

【摘要】 由于彩色图像表示的三帧之间存在着色度冗余,如果不针对彩色图像表示的性质来充分利用色度冗余信息,那么将不会形成更为有效的压缩编码。与传统的彩色图像的压缩方法相比,基于MPEG-21的三维矩阵理论能够将彩色图像用一个统一的数学模型来表示,通过RGB空间到YCbCr空间的颜色空间转换后,采用新的三维子阵联合分割方式,将分块后的两类三维子阵作为一个数字项来考虑,能够与未来多媒体框架标准MPEG-21有很好的兼容性;同时采用线性非均匀标量量化方法,进一步提高了编码效率。仿真实验结果表明该方法的性能要稍优于JPEG标准。利用三维矩阵变换对彩色图像压缩编码的有效性,本文从主观和客观两方面出发,给出了一种来更为符合人眼视觉感知特性的彩色图像质量综合评价方案。更多还原

【Abstract】 With the development of Internet and multimedia application technologies, color images are used widely than ever, for which can reflect the real world and comply with the perception of human eyes, additionally, the advance of compression and coding technologies for color images is one of key factors. An uncompressed color image will need more space, which will cause more trouble for its transmission over Internet; however, compressed one will need smaller space and can be transmitted conveniently.It is well known that the fundamental theory is that there is much redundant information within one image according to information theory and generally speaking, the data distribution rule of one image cannot be observed only by human eyes and there is much relevance between image data representation by statistics, such as the position relevance of pixels. For a color image represented by RGB (Red, Green and Blue) color space, there exists not only position relevance for pixels in one frame but also color relevance in the same position of different frames, i.e. redundant information in one color image includes statistic, structural, knowledge and visional redundancy as well as color redundancy, which also indicates the specificity of compression and coding for color image and usually, the method of color space conversion is used firstly to remove some color redundancy and then the compression methods adopted for gray image is transplanted directly.Because image is two-dimensional data representation and the progress of data compression technologies will also accelerate that of image compression technologies, transforms, such as DCT and wavelet transform, are used to convert image representation from space domain to frequency or transformed domain and correspondingly, many compression methods, such as RLE (Run-Length Encoding), predictive coding, Huffman coding and arithmetic coding will also be更多还原