數(shù)字圖像處理模擬題(雙語課)(共10頁)

1、Terms and concepts explanationRGB：Red Green Blue 三原色 紅 綠 藍CMYK：cyan magenta yellow black 青、粉紅(fnhng)、黃、黑HSI: hue saturation intensity 色調(diào)(s dio)、飽和度、亮度FFT fast fourier transform 快速(kui s)傅里葉變換CWT continuous wavelet transform 連續(xù)小波變換DCT discrete cosine transform 離散余弦變換DFT discrete fourier transform 離散傅

2、里葉變換DWT discrete wavelet transform 離散小波變換CCD charge-coupled device電荷耦合元件Pixel a digital image is composed of a finite number of elements,each of which has a particular lication and value,these elements are called pixel 像素DC component in frequency domain （direct currentcomponent）直流分量的頻率域GLH The Gray

3、Level Histogram 灰度直方圖Mather（basic） wavelet :a function (wave) used to generate a set of wavelets,母小波，用于產(chǎn)生小波變換所需的一序列子小波Basis functions basis image : there is only one set of k for any given f(x), then the k (x) are called basis functionsMulti-scale analysis 多尺度分析Gaussian function :Gaussian functionIn

4、 mathematics,is a function of the form: for some real constants a 0, b, c 0, and e 2.718281828 (Eulers number).對于一些真正的常量0,b,c 0,和e2.718281828(歐拉數(shù))。Sharpening filter ：銳化濾波器Smoothing filter/convolution ：smoothingfilterareusedforblurringandfornoisereduction平滑濾波器用于模糊處理和降低噪聲/卷積Imageenhancement/imageresto

5、rationimageEnhancementtheprocessofmanipulatinganimagesothattheresultismoresuitablethantheoriginalforaspecificapplication.圖像增強處理是對圖像進行的加工，使其結果對于特定的應用比原始圖像更適合的一種處理。Imagerestorationisanareathatalsodealswithimprovingtheappearanceofanimage.However,unlikeenhancement,whichissubjective,imagerestorationisobj

6、ective.圖像復原也是一個改進圖像外觀的一個處理領域，然而與圖像增強不同，圖像增強是主觀的，圖像復原是客觀的。取樣(qyng)sampling: Digitizing the coordinate values is called sampling.對坐標值進行數(shù)字化稱為.量話quantization：Digitizing the amplitude values is called quantization.對幅值數(shù)字化稱為(chn wi).空間分辨率：spatial resolution is a measure of the smallest discernible detail i

7、n an image.圖像中可辨別的最小細節(jié)(xji)的度量灰度分辨率 ：Intensity resolution refers to the smallest discernible change in intensity level.灰度分辨率是指在灰度級中可分辨的最小變化空間域濾波：Spatial domain filteringFrequency domain filtering：Frequency domain filtering with a variable frequency for the signal filtering 頻率域濾波以頻率作為變量對信號進行濾波wavelet

8、 transforms are based on small waves,of varying frequency and limited duration.小波變換基于一些小型波，它具有變化的頻率和有限的持續(xù)時間。Image compression ,the art and science of reducing the amount of data required to represent an image.圖像壓縮是一種減少描繪一幅圖像所需數(shù)據(jù)量的技術和科學. Cite one example of digital image processing？Answer: In the dom

9、ain of medical image processing we may need to inspect a certain class of images generated by an electron microscope to eliminate bright, isolated dots that are no interest. Cite one example of spatial operation舉一個空間操作的例子Answer: In the domain of medical image processing we may need to inspect a cert

10、ain class of images generated by an electron microscope to eliminate bright, isolated dots that are no interest. From the following processing result make a general comment about ideal highpass(figure B)and Gaussian highpass filter(figure D) Answer：Gaussian highpass filter is more smooth than ideal

11、highpass, even for small objects and thin lines with GHPF filter is also more clearThe original image, the ideal lowpass filter and Gaussian lowpass filter are shown beB and C D and E are the results of the either filter B or CDraw lines to connect the filters with their result 連線(lin xin)濾波器和處理結果Ex

12、plain the difference of the two filters 解釋(jish)兩者的區(qū)別Answer：It is clear from this example that ideal lowpass filtering is not very practical實用(shyng),it has ringing properties. Over characteristics of Gauss filter is very flat, so it is not ringingBlurred with a 3x3 smoothing mask would the resultan

13、t histograms still be the same? Draw the two histograms and explain your answer模糊了一個3x3的平滑的掩膜？會得到的直方圖仍然是相同的嗎？畫兩圖并解釋你的答案With the chromaticity diagram bellow give a brief description to the RGB color model. And these three colors enough to compose all visible colors?用色度圖給出的RGB顏色模型的簡短描述。這三種顏色足以構成所有可見的顏

14、色嗎？Answer：Images represented in the RGB color model consist of three component images, one for each primary color.These three colors enough to compose all visible colorsWhat the result when applying an averaging mask with the size 1x1？什么樣的結果時，應用平均掩膜大小1x1？不變 A mean filter is a linear filter but a med

15、ian filter is not. WHY？均值濾波器是一種線性濾波器，中值濾波器不是，為什么？The basic principle of linear filtering is to replace the original image with the mean value of each pixel, but median filter replace the original image with the median value of each pixel.The value of mean and median is different.Develop an algorithm

16、 which implements frequency domain filtering by means of Fourier transform. 利用傅立葉變換實現(xiàn)了頻率(pnl)域濾波的算法。Answer：The steps of the process are as follow:(1) Multiply the input image f(x,y) by （-1）x+y to center the transform; （1）將輸入圖像f（x，y）的（-1）x+y為中心(zhngxn)的變換；(2) Compute the DFT of the image from (1) to

17、get power spectrum F(u,v) of Fourier transform.計算圖像(t xin)的DFT從（1）得到的功率譜f（u，v）的傅里葉變換。Multiply by a filter function h(u,v) 乘以一個濾波器函數(shù)h(u,v) .Compute the inverse DFT of the result in (3).Obtain the real part of the result in (4).Multiply the result in (5) by（-1）x+y The following matrix A is a 3x3 image

18、 and B is a 3x3 laplacian mask, what will be the resulting image?(Note that the elements beyond the border remain unchanged)Develop an algorithm to obtain the processing result B from original image ADevelop an algorithm which computes the pseudo-color image by means of fourier transformAnswer：The s

19、teps of the process are as follow:(1) Multiply the input image f(x,y) by （-1）x+y to center the transform; (2) Compute the DFT of the image from (1) to get power spectrum F(u,v) of Fourier transform.(3) Multiply by a filter function h(u,v) .(4) Compute the inverse DFT of the result in (3).(5) Obtain

20、the real part of the result in (4).(6) Multiply the result in (5) by（-1）x+y （頻率域的基本濾波步驟）Digital image histogram 編寫一個函數(shù)(hnsh)計算數(shù)字圖像直方圖；1)創(chuàng)建(chungjin)函數(shù) h：function h=hist(x,N) 2)算法(sun f)代碼：（Untitled6.M）m,n=size(x); a=imread(F:Border Collie.jpg);h=zeros(1,N); a=rgb2gray(a);x=double(x); b=hist(a,256);f

21、or i=1:m bar(b); for j=1:n figure,imhist(a); h(x(i,j)+1)=h(x(i,j)+1)+1; endendFundamentalStepsinimagesDigitalimageProcessing數(shù)字圖像圖像處理的基本步驟imageacquisition圖像采集image filter andenhancement圖像濾波和增強imagerestoration圖像復原Colorimageprocessing彩色圖像處理wavelets and multiresolution processing小波與多分辨率處理compression（壓縮）

22、morphologicalprocessing(形態(tài)學處理)segmentation（分割）representationanddescription（表示與描述）object recognition（目標識別）Lossless approachesHoffman Coding 無損方法 - 霍夫曼編碼步驟： create of source reductions by ordering the symbols under consideration and combining the lowest probability symbols into a single symbols that r

23、eplaces them in the next source reduction.Code each reduced source, starting with the smallest source and working back to the original source.Simple programming with MATLAB1.There are two satellite photos of night as below. Write a program with MATLAB to tell which is brighter有衛(wèi)星照片的夜晚如下。寫一個程序用MATLAB

24、來得知(d zh)哪個照片是更亮的代碼(di m)：A=imread（1.jgp）; B=imread(2.jpg);m,n=size(A);for i=1:mfor j=1:n sum1=sum1+AI,j;endendavg1=sum1/m*n;r,c=size(B);for i=1:mfor j=1:n sum2=sum2+BI,j;endendavg2=sum2/m*n;An 8x8 images f(i,j) has gray levels given by the following equationF(i,j)=|i-j| i,j=0,1 Write a program to f

25、ind the output image obtained by applying a 3x3 median filter on the image f(i,j);note that the border pixels remain unchanged有一個8x8圖像f（i，j）灰度由以下(yxi)公式得出F（i，j）= | I-J | i，j = 0,1 .寫一個程序(chngx)來得出用 中值 濾波器采用3x3圖像f（i，j）的輸出圖像；注意(zh y)邊界像素保持不變Avaege filter均值濾波function r=avgfilter(gray,n) a(1:n,1:n)=1;ro

26、w,col=size(gray);gray1=double(gray); gray2=gray1; for i=1:row-n+1 for j=1:col-n+1 c=gray1(i:i+(n-1),j:j+(n-1).*a; s=sum(sum(c); gray2(i+(n-1)/2,j+(n-1)/2)=s/(n*n); end end r=uint8(gray2); avg3=avgfilter(noise,3); avg5=avgfilter(noise,5); avg7=avgfilter(noise,7); subplot(221);imshow(noise);title(原噪聲圖

27、); subplot(222);imshow(avg3);title(3*3均值濾波圖); subplot(223);imshow(avg5);title(5*5均值濾波圖); subplot(224);imshow(avg7);title(7*7均值濾波圖);Winener filrer維納濾波 figure;subplot(2,3,1);imshow(gray);title(原灰度圖像); noise2 = imnoise(gray,gaussian,0,0.05); subplot(2,3,3);imshow(noise2);title(高噪點處理后的圖像); winener3=wien

28、er2(noise2,3 3); winener5=wiener2(noise2,5 5); winener9=wiener2(noise2,9 9); subplot(2,3,4);imshow(winener3);title(維納濾波器處理后的圖像3*3); subplot(2,3,5);imshow(winener5);title(維納濾波器處理后的圖像5*5); subplot(2,3,6);imshow(winener9);title(維納濾波器處理后的圖像9*9);median filter中值濾波function r=midfilter(gray,n) row,col=size(

29、gray);gray1=double(gray); gray2=gray1; for i=1:row-n+1 for j=1:col-n+1 c=gray1(i:i+(n-1),j:j+(n-1); e=c(1,:); for u=2:n e=e,c(u,:); end mm=median(e); gray2(i+(n-1)/2,j+(n-1)/2)=mm; end end r=uint8(gray2); mid3=midfilter(noise,3); mid5=midfilter(noise,5); mid7=midfilter(noise,7); subplot(221);imshow(

30、noise);title(原噪聲(zoshng)圖); subplot(222);imshow(mid3);title(3*3中值濾波(lb)圖); subplot(223);imshow(mid5);title(5*5中值濾波(lb)圖); subplot(224);imshow(mid7);title(7*7中值濾波圖);Geometric mean filter幾何均值濾波function r=geometric(gray,n) a(1:n,1:n)=1; hight, width=size(gray); x1=double(gray); x2=x1; for i=1:hight-n+1

31、 for j=1:width-n+1 c=x1(i:i+(n-1),j:j+(n-1).*a; s=prod(prod(c); x2(i+(n-1)/2,j+(n-1)/2)=nthroot(s, n*n); end end r=uint8(x2); img=imread(E:university數(shù)字圖像處理實驗4test.jpg); gray=togray(img); subplot(2,3,1);imshow(gray);title(原灰度圖像); noise = imnoise(gray,gaussian,0,0.02); subplot(2,3,3);imshow(noise);tit

32、le(高噪點處理后的圖像); adaptgray=adaptive(noise,3); subplot(2,3,4);imshow(adaptgray);title(自適應(shyng)局部降噪濾波器); avg=avgfilter(noise,3); subplot(2,3,5);imshow(avg);title(3*3算術(sunsh)均值濾波圖像); geo=geometric(noise,3); subplot(2,3,6);imshow(geo);title(3*3幾何均值(jn zh)濾波圖像);The theory of the Wiener filterThe Wiener filtering is optimal in terms of the mean square error. In other words, it minimizes the overall mean square error in the process of inver