节点文献
3D-DCT自适应彩色图像压缩算法的研究
Adaptive Color Image Compressiong Algorithm Based on3D-DCT
【作者】 宋琎琦；
【导师】 刘彦隆；
【作者基本信息】 太原理工大学 ， 电子与通信工程， 2012， 硕士
【摘要】 由于数字图像信息给人以直观具体的立体彩色形象,并且具有传输质量好、可靠性高、易于加密等优点,因而广泛应用于人们的日常生活中。但随着人们对清晰度要求的不断提高,数字图像的信息量不断增大,消耗了大量的存储空间和传输带宽。因此,在保持一定编码质量的前提下,如何减少信息量以达到数据压缩是一个值得研究的课题。离散余弦变换(DCT)是一种较好的基于频率域的压缩算法,该算法主要依据的原理是：在一幅图像中,其直流和低频分量占主要部分,故可通过舍去部分次要的高频分量来达到压缩的目的。又因该算法具有极为接近K-L变换的去相关性并且有快速算法,所以被各图像及视频压缩的国际标准广泛采用。随着计算机处理能力的提高,使早期提出的三维离散余弦变换(3D-DCT)的应用成为可能。3D-DCT技术就是通过进一步压缩彩色图像中因各颜色分量所具有相同纹理,边缘和灰度变化梯度所产生的冗余,从而达到提高图像压缩的目的。本文提出了一种以峰值信噪比(PSNR)和压缩比(CR)为参考标准的基于3D-DCT的自适应图像压缩算法。该算法主要提出了一种与二值编码器(TRE)相结合的新的量化方法,并将其应用在对重构图像质量影响最大的几个三维离散余弦变换后的系数上,以达到提高CR的目的。本文最后使用MATLAB软件对所提出的算法进行了仿真并与当前流行的JPEG编码进行比较,证明文中算法在追求较高CR的情况下也保证了较好的PSNR。
【Abstract】 As the digital image information provides visualize specific three-dimensional color image and has some advantages such as good transmission quality, high reliability and easy encryption, it is widely used in people’s daily life. However, with the rising requirements of digital image definition, the quantity of digital image information is increasing ceaselessly, which consumes a large amount of storage space and transmission bandwidth. Therefore, how to reduce the amount of information to achieve data compression while maintaining a certain encoding quality becomes a topic worthy of study.Discrete cosine transform (DCT) is a better compression algorithm based on frequency domain, since it can eliminate by the high frequency component part of secondary image to achieve the aim of compression while fully preserving the DC and low frequency components which are the main part of the image. The decorrelation of the transform is close to the optimal K-L transform and there is a fast algorithm for DCT, so DCT is widely adopted by many international image and video compression standards. With the enhancement of computer processing ability, the application of three-dimensional discrete cosine transform(3D-DCT) which was put forward early becomes possible.3D-DCT technology can eliminate by the redundancy of the similar texture, edge and gray level gradient to reach the high compression ratio.In this paper, an adaptive color image compression algorithm based on three-dimensional discrete cosine transform is proposed. The algorithm use PSNR and CR as its reference standard. For the aim to obtain the best possible compression ratio CR, a new quantitative method combined with two role encoder(TRE) is proposed and applies to some coefficients of three-dimensional discrete cosine transform which have more impact on reconstructed image quality. Finally this paper uses MATLAB for the simulation. With the popular JPEG coding for comparison, we find that the algorithm makes a high compression ratio and also ensures a better PSNR.