毕业论文外文翻译-小波在图像压缩领域中的应用.doc

《毕业论文外文翻译-小波在图像压缩领域中的应用.doc》由会员分享，可在线阅读，更多相关《毕业论文外文翻译-小波在图像压缩领域中的应用.doc（11页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、河北大学工商学院2015届本科生毕业外文翻译摘要小波分析提供了一种强大的和非常灵活的工具集，来处理在科学和工程方面的基本问题，比如音频消噪、信号压缩、目标探测和指纹压缩、图像去噪、图像增强、图像识别、诊断心脏病和语音识别等等。本文主要介绍小波在图像压缩领域中的应用，包括实现过程、数学推导以及结果分析。小波图像压缩是用各种已知的具有不同数学特性的小波来完成的。我们研究数学小波应用的见解方式是为了符合图像压缩的工程模型。1介绍小波函数根据尺度或分辨率进行信号或图像的数据分析。信号处理的小波算法事实上和人类眼睛的工作方式很像;也和数码相机处理视觉缩放分辨率及中间细节的方式很像。同样的原则它也捕捉手机

2、信号,及数字化彩色图像在医学上的使用。小波分析的真正应用领域, 例如在具有明显的间接性如易变方向的信号的数据，或拥有大量边缘的图片。而小波也许是函数理论中的一章，我们证明这个结果的关键是数字处理算法，或更精确的数字化信息，信号，时间序列，图像，电影等。因此，小波思想的应用包括大部分的信号和图像处理，数据压缩，指纹编码，和许多其它科学和工程领域。本论文重点是使用定制设计的小波算法,和数学阈值过滤器对彩色图像的处理。2彩色图像小波压缩小波图像压缩的全过程执行如下: 由计算机输入图像，提出小波变换对数字图像进行阈值分割，对数字图像进行熵编码，在必要的地方完成图像，因此，图像压缩是在计算机上完成的。完

3、成压缩图像、小波转换图像重建,然后在图像上进行逆小波变换,因此图像重建。在某些情况下,图像算法(Sha93)的使用是已知的较好的压缩图像的算法,但它不能在这里实现2.1提出小波变换各种小波变换用于这一步。即Daubechies小波，Coiflets,双正交小波,Symlets。这些不同的转换是观察各种数学性质,如对称,号码消失的时刻和正交如何实现压缩图像的不同结果。优势短支持保留位置。使用的Daubechies小波,Coiflets是正交的。 Symlets的性能接近对称。双正交小波不正交也没有在各种过滤器上正交从而允许他们拥有对称属性。MATLAB有一称为wavedec2的子程序,它执行图像

4、的分解以达到给定的期望水平(N)与给定的期望小波(wname)。这有三个组件来处理小波变换应用于分量方式。“wavedec”是一个二维小波分析功能。C,S= wavedec2(X,N,wname)返回矩阵X N级小波分解,用小波在字符串“wname”中命名。输出是分解矢量C和相应的书中保持矩阵SMatlabUG。这里的图像作为矩阵X。2.2阈值这个项目的目的是使用固定阈值比较每一个使用不同小波图像压缩的性能。MATLAB这个子程序称为wthrmngr，它根据选项和方法计算全局阈值或水平依赖阈值。可用的选项有全局阈值和水平依赖阈值,及用于程序的全球阈值。然而,使用一个固定的阈值,是为了每一个小波

5、变换具有相同的给定条件来比较不同条件下的性能。这个项目使用的固定阈值是10到20。用无损压缩0作为阈值,原因很明显。2.3 小波 紧支撑小波函数定义在一个有限区间且平均值为零序列中。小波变换的基本思想是代表任意函数f(x)等叠加的一组小波基函数。这些基础功能通过扩张或缩放及译文从aingle原型获得小波被称为母小波(x)。除了少量的间断，小波基非常擅长有效地代表平滑函数。3图像的数学表示法3.1数字图像表示法一个图像被定义为一个二维函数。一个矩阵f(x,y),x和y是空间坐标, ,f的振幅是任何一对坐标(x,y)称为强度或灰度图像的点。彩色图像相结合形成单独的二维图像。例如,在RGB颜色系统，

6、彩色图像由三个即红色,绿色和蓝色的单个组件组成的图像。因此许多技术开发是为了单色图像可以通过处理三个单独的组件图片扩展到彩色图像。当x,y和f的振幅值都是有限的,离散的数量,图像被称为数字图像。数字图像处理领域是指通过数字计算机处理数字图像。数字图像是由有限数目的元素组成,每一种都有一个特定的位置和幅值。这些元素被称为图像元素,图像像素,图元和像素。由于像素是最广泛使用的术语,这些元素从现在开始将被表示为像素。图像在x和y坐标上,及振幅上可能是连续的。数字化坐标以及振幅能够使图像转换到数字形式。在这坐标值称为采样的数字化,数字化的振幅值称为量化。数字图像是由有限数目的元素组成,每一种都有一个特

7、定的位置和幅值。数字图像处理领域是指通过数字计算机处理的数字图像。3.2MATLAB里读取图片图像使用名为imread的函数读取到MATLAB环境中。语法如下:imread(文件名),文件名包含完整的任何适用扩展字符的图像文件名称。例如,命令行 f = imread(lena.jpg);莉娜JPEG图像读入图像数组或图像矩阵f。图像中由于有三个颜色组件,即红色,绿色和蓝色的成分,图像被分解成三个不同的颜色矩阵fR fG和facebook。3.3 图像的小波分解颜色转换，图像压缩的过程中,应用压缩图像的RGB组件会导致不良的颜色变化。因此,图像转化为强度、色调和色彩饱和度组件。JPEG 2000

8、标准中使用的颜色转换Sko01已经采用。对于有损压缩,方程(3.2)和(3.3)用于程序。4结果和讨论4.1 讨论有损压缩。有多种因素影响图像压缩。在第二节如上所述,小波的非正交性可能导致有损压缩。应用压缩阈值时,一些微不足道的系数扔掉也会导致有损压缩。同时,应用小波变换的水平数量会影响压缩质量。当某些小波使用时lossiness造成的非正交小波是不可以避免的, 应用小波变换是为了减少lossiness，它通过一直降低单一像素级来降低色阶数量 (floor(log2(min(size of Image)。除此之外应用各种阈值来研究lossiness。有损压缩方法往往会产生错误的解压缩图像。有损

9、压缩方法在误差极小,甚至无法察觉时使用。如果那些无法察觉的误差是可以接受的，有损技术相较于无损技术在达到更高的压缩比方面更有优势。为了支持成果图像的压缩和理论知识这个主张,我们从文本获得一些数值比较。它们的压缩率,均方根误差,均方根,相对两个标准差异,D,和峰值信噪比,PNSR。使用的公式如下: 有两种不同阈值的各种小波变换被用来压缩lena.png 8位彩色图像。结果如下: 一件事可以马上指出通过查看图像,图像压缩与较小的阈值是10相比与压缩阈值20看起来更接近原始的丽娜图像。4.2结论小波压缩展现了显著的性能，尤其是使用较小的阈值时, 一些情况下在原始图像和压缩图像之间是不可区分的。然而,

10、仍然可以做出更多的改进。因为它在Sko01中提到通过添加更多的压缩等量化熵编码阶段,有更多进步的空间通过。同时,我们还没有覆盖所有的小波,它不能决定哪个是表现最好的图像压缩。AbstractWavelets provide a powerful and remarkably flexible set of tools for handling fundamental problems in science and engineering, such as audio de-noising, signal compression, object detection and fingerprint

11、 compression, image de-noising, image enhancement, image recognition, diagnostic heart trouble and speech recognition, to name a few. Here, we are going to concentrate on wavelet application in the field of Image Compression so as to observe how wavelet is implemented to be applied to an image in th

12、e process of compression, and also how mathematical aspects of wavelet affect the compression process and the results of it. Wavelet image compression is performed with various known wavelets with different mathematical properties. We study the insights of how wavelets in mathematics are implemented

13、 in a way to fit the engineering model of image compression.1. IntroductionWavelets are functions which allow data analysis of signals or images, according to scales or resolutions. The processing of signals by wavelet algorithms in factworks much the same way the human eye does; or the way a digita

14、l camera processes visual scales of resolutions, and intermediate details. But the same principle also captures cell phone signals, and even digitized color images used in medicine.Wavelets are of real use in these areas, for example in approximating data with sharp discontinuities such as choppy si

15、gnals, or pictures with lots of edges.While wavelets is perhaps a chapter in function theory, we show that the algorithms that result are key to the processing of numbers, or more precisely of digitized information, signals, time series, movies, color images, etc. Thus, applications of the wavelet i

16、dea include big parts of signal and image pro-cessing, data compression, fingerprint encoding, and many other fields of science and engineering. This thesis focuses on the processing of color images with the use of custom designed wavelet algorithms, and mathematical threshold filters.2Wavelet Color

17、 Image CompressionThe whole process of wavelet image compression is performed as follows: An input image is taken by the computer, forward wavelet transform is performed on the digital image, thresholding is done on the digital image, entropy coding is done on the image where necessary, thus the com

18、pression of image is done on the computer. Then with the compressed image, reconstruction of wavelet transformed image is done, then inverse wavelet transform is performed on the image, thus image is reconstructed. In some cases, zero-tree algorithm Sha93 is used and it is known to have better compr

19、ession with zero-tree algorithm but it was not implemented here.2.1 Forward Wavelet Transform. Various wavelet transforms are used in this step. Namely, Daubechies wavelets, Coiflets, biorthogonal wavelets, and Symlets. These various transforms are implemented to observe how various mathematical pro

20、perties such as symmetry, number of vanishing moments and orthogonality differ the result of compressed image. Advantages short support is that it preserves locality. The Daubechies wavelets used are orthogonal, so do Coiflets. Symlets have the property of being close to symmetric. The biorthogonal

21、wavelets are not orthogonal but not having to be orthogonal gives more options to a variety of filters such as symmetric filters thus allowing them to possess the symmetric property.MATLAB has a subroutine called wavedec2 which performs the decomposition of the image for you up to the given desired

22、level (N) with the given desired wavelet(wname). Since there are three components to deal with, the wavelet transform was applied componentwise. “wavedec” is a two-dimensional wavelet analysis function. C,S = wavedec2(X,N,wname) returns the wavelet decomposition of the matrix X at level N, using the

23、 wavelet named in string wname. Out-puts are the decomposition vector C and the corresponding book keeping matrix SMatlabUG. Here the image is taken as the matrix X.2.2 Thresholding.Since the whole purpose of this project was to compare the performance of each image compression using different wavel

24、ets, fixed thresholds were used.MATLAB has this subroutine called wthrmngr which computes the global threshold or level dependent thresholds depending on the option and method. The options available are global threshold and level dependent threshold, and the global threshold is used in the program.

25、However, a fixed threshold values were used so as to have the same given condition for every wavelet transform to compare the performances of different conditions. Here, fixed thresholds 10 and 20 were used.For the lossless compression 0 is used as the threshold for an obvious reason.2.3WaveletsComp

26、actly supported wavelets are functions defined over a finite interval and having an average value of zero. The basic idea of the wavelet transform is to represent any arbitrary function f(x) as a superposition of a set of such wavelets or basis functions. These basis functions are obtained from aing

27、le prototype wavelet called the mother wavelet (x), by dilations or scaling and translations. Wavelet bases are very good at efficiently representing functions that are smooth except for a small set of discontinuities.3Digital Image Representation and Mathematics behind It3.1 Digital Image Represent

28、ation. An image is defined as a two-dimensional function . a matrix, f(x, y), where x and y are spatial coordinates, and the amplitude of f at any pair of coordinates (x, y) is called the intensity or gray level of the image at the point. Color images are formed by combining the individual two-dimen

29、sional images. For example, in the RGB color system, a color images consists of three namely, red, green and blue individual component images. Thus many of the techniques developed for monochrome images can be extended to color images by processing the three component images individually. When x, y

30、and the amplitude values of f are all finite, discrete quantities, the image is called a digital image. The field of digital image processing refers to processing digital images by means of a digital computer. A digital image is composed of a finite number of elements, each of which has a particular

31、 location and value. These elements are referred to as picture elements, image elements, pels and pixels. Since pixel is the most widely used term, the elements will be denoted as pixels from now on.An image maybe continuous with respect to the x- and y-coordinates, and also in amplitude. Digitizing

32、 the coordinates as well as the amplitude will take into effect the conversion of such an image to digital form. Here, the digitization of the coordinate values are called sampling; digitizing the amplitude values is called quantization. A digital image is composed of a finite number of elements, ea

33、ch of which has a particular location and value. The field of digital image processing refers to processing digital images by means of a digital computer.3.2. Reading Images. In MATLAB.images are read into the MATLAB environment using function called imread. The syntax is as follows: imread(filename

34、) Here, filename is a string containing the complete name of the image file including any applicable extension. For example, the command line f = imread (lena.jpg); reads the JPEG image lena into image array or image matrix f. Since there are three color components in the image, namely red, green an

35、d blue components, the image is broken down into the three distinct color matrices fR fG and fB。3.3. Wavelet Decomposition of an Image.Color Conversion. In the process of image compression, applying the compression to the RGB components of the image would result in undesirable color changes. Thus, t

36、he image is transformed into its intensity, hue and color saturation components. The color transformation used in JPEG 2000 standard Sko01 has been adopted. For the lossy compression, equations (3.2) and (3.3) were used inthe program.4. Results and Discussion4.1. Discussion.Lossy Compression. There

37、are various factors that influence the image compression. As mentioned above in section 2, nonorthogonality of the wavelet may cause the compression to be lossy. When threshold is applied to the compression, some of the insignificant coefficients are thrown out thus the resulting in lossy compressio

38、n. Also, the number of levels the wavelet transform is applied would influence the compression quality. Although the lossiness caused by the nonorthogonal wavelet was not avoidable when certain wavelets were used, an attempt to minimize the lossiness was made for the number of levels part by going d

39、own all the way to the single pixel level when the wavelet transform was applied (floor(log2(min(size of Image). In addition various threshold values are appliedto observe the lossiness. A lossy compression method tend to produce inaccuracies in the decompressed image. Lossy compression method is us

40、ed when these inaccuracies are so small that they are imperceptible. If those imperceptible inaccuracies are acceptable the lossy technique is advantageous compared to the lossless ones for higher compression ratios can be attained.In order to support the claims made by comparison of the resulting i

41、mages and the theoretical knowledge that we obtained from the texts, some numerical comparisons are made. They are the compression ratio, the root mean square error, rms, the relative two norm difference, D, and the peak signal to noise ratio,PNSR. The formulas used are as follows:Various wavelet tr

42、ansforms with two different thresholdings were used to compress the and 8-bit color image lena.png. The results are as follows:One thing that could be noted right away by looking at the images is that the images compressed with smaller threshold value that is 10 look closer to the original lena.png

43、compared to the images compressed with threshold value 20 overall.4.2. Conclusion.Wavelet compression did show remarkable performance especially with smaller threshold value; it was not differentiable in between the original image and the compressed image for some cases.However, more improvements ca

44、n still be made. As it is mentioned in Sko01 there is more room for improvement by adding more stages to the compression such as quantization, entropy encoding, etc. Also, we have not covered all the wavelets that is out there, that it cannot be decided as to which one performs the best image compre

45、ssion.五分钟搞定5000字毕业论文外文翻译，你想要的工具都在这里!在科研过程中阅读翻译外文文献是一个非常重要的环节，许多领域高水平的文献都是外文文献，借鉴一些外文文献翻译的经验是非常必要的。由于特殊原因我翻译外文文献的机会比较多，慢慢地就发现了外文文献翻译过程中的三大利器：Google“翻译”频道、金山词霸（完整版本）和CNKI“翻译助手。具体操作过程如下： 1.先打开金山词霸自动取词功能，然后阅读文献； 2.遇到无法理解的长句时，可以交给Google处理，处理后的结果猛一看，不堪入目，可是经过大脑的再处理后句子的意思基本就明了了； 3.如果通过Google仍然无法理解，感觉就是不同，那

46、肯定是对其中某个“常用单词”理解有误，因为某些单词看似很简单，但是在文献中有特殊的意思，这时就可以通过CNKI的“翻译助手”来查询相关单词的意思，由于CNKI的单词意思都是来源与大量的文献，所以它的吻合率很高。 另外，在翻译过程中最好以“段落”或者“长句”作为翻译的基本单位，这样才不会造成“只见树木，不见森林”的误导。四大工具： 1、Google翻译： google，众所周知，谷歌里面的英文文献和资料还算是比较详实的。我利用它是这样的。一方面可以用它查询英文论文，当然这方面的帖子很多，大家可以搜索，在此不赘述。回到我自己说的翻译上来。下面给大家举个例子来说明如何用吧比如说“电磁感应透明效应”这

47、个词汇你不知道他怎么翻译，首先你可以在CNKI里查中文的，根据它们的关键词中英文对照来做，一般比较准确。 在此主要是说在google里怎么知道这个翻译意思。大家应该都有词典吧，按中国人的办法，把一个一个词分着查出来，敲到google里，你的这种翻译一般不太准，当然你需要验证是否准确了，这下看着吧，把你的那支离破碎的翻译在google里搜索，你能看到许多相关的文献或资料，大家都不是笨蛋，看看，也就能找到最精确的翻译了，纯西式的！我就是这么用的。 2、CNKI翻译： CNKI翻译助手，这个网站不需要介绍太多，可能有些人也知道的。主要说说它的有点，你进去看看就能发现：搜索的肯定是专业词汇，而且它翻译

48、结果下面有文章与之对应（因为它是CNKI检索提供的，它的翻译是从文献里抽出来的），很实用的一个网站。估计别的写文章的人不是傻子吧，它们的东西我们可以直接拿来用，当然省事了。网址告诉大家，有兴趣的进去看看，你们就会发现其乐无穷！还是很值得用的。 3、网路版金山词霸（不到1M）： 4、有道在线翻译：翻译时的速度：这里我谈的是电子版和打印版的翻译速度，按个人翻译速度看，打印版的快些，因为看电子版本一是费眼睛，二是如果我们用电脑，可能还经常时不时玩点游戏，或者整点别的，导致最终SPPEED变慢，再之电脑上一些词典（金山词霸等）在专业翻译方面也不是特别好，所以翻译效果不佳。在此本人建议大家购买清华大学编写的好像是国防工业出版社的那本英汉科学技术词典，基本上挺好用。再加上网站如：google CNKI翻译助手，这样我们的翻译速度会提高不少。具体翻译时的一些技巧（主要是写论文和看论文方面） 大家大概都应预先清楚明白自己专业方向的国内牛人，在这里我强烈建议大家仔细看完这些头上长角的人物的中英文文章，这对你在专业方向的英文和中文互译水平提高有很大帮