gpt-3.5-turbo-0301· openai
Suzhen/CodeChat
hello
- User
hello
- Assistant
Hello there! How can I assist you today?
- User
我想要写一个matlab程序
- Assistant
非常好!请告诉我您需要编写的 MATLAB 程序的目的和功能。我会尽力帮助您开始编写程序。
- User
我需要用matlab完成灰度图像降噪功能,使用gdf方法
- Assistant
好的,以下是一个使用GDF方法进行灰度图像降噪的 MATLAB 示例程序:
% 读入图像 I = imread('lena.jpg'); figure subplot(121),imshow(I),title(' Original image '); % Gaussian滤波参数设置 Gsigma=2.0; Gsize=5; % 图像灰度化 Igray=double(I(:,:,1))*0.2989+double(I(:,:,2))*0.5870+double(I(:,:,3))*0.1140; subplot(122),imshow(uint8(Igray)),title('Gray level image'); % GDF 方法降噪 Ksigma=1.8; Ksize=39; Iblur=Gaussianfilter(Igray,Gsize,Gsigma); figure,imshow(Iblur,[]),title('Gaussian filter result'); K = GDF(Iblur,Ksize,Ksigma); Iout = max(0, Iblur - K); figure,imshow(Iout,[]),title('GDF result'); % 高斯滤波函数 function G = Gaussianfilter(I,size,sigma) [X,Y] = meshgrid(-floor(size/2):floor(size/2),-floor(size/2):floor(size/2)); dist = (X.^2+Y.^2)/(2*sigma*sigma); G = exp(-dist)/(2*pi*sigma*sigma); G = G/sum(sum(G)); I=im2double(I); G=im2double(G); if (length(size(I))>2) G=repmat(G,[1 1 size(I,3)]); for i=1:size(I,3) G(:,:,i)=fft2(G(:,:,i)); I(:,:,i)=fft2(I(:,:,i)); I(:,:,i)=I(:,:,i).*G(:,:,i); I(:,:,i)=ifft2(I(:,:,i)); end else G=fft2(G); I=fft2(I); I=I.*G; I=ifft2(I); end G=real(I); end % GDF 函数 function K=GDF(I,size,sigma) dsize=round(size/2)-1; [X,Y]=meshgrid(-dsize:dsize,-dsize:dsize); m=-exp(-((X.^2+Y.^2)/(2*(sigma^2)))); m=dsize+1-m; K=zeros(size(size(I))); for i=dsize+1:(size(I,1)-dsize) for j=dsize+1:(size(I,2)-dsize) win=I(i-dsize:i+dsize,j-dsize:j+dsize); wm=winsor(m, win); % 计算 K K(i,j)=sum(sum(wm.*(win-ones(size(wm)).*median2(win)))); end end K=K/(size^2); end % Winsor 函数 function W=winsor(m, win) s(1)=sum(sum(m>=0))*2-1; s(2)=sum(sum(m<0)); msorted=sort(m(m~=0)); msorted=[-flipud(msorted(:)); msorted(:)]; ms=msorted.*(msorted<0); del=max(s([1 2]),1); ms=ms(del:end-del+1); W=win; for i=1:length(m(:)) if (m(i)~=0) r=max(1,i-size(W,1)*(size(W,2)-1)/2); c=i-(r-1)*size(W,1); linw=reshape(W(floor(max(1,r-ceil(size(W,1)/2))):min(end,ceil(r+size(W,1)/2)-1),floor(max(1,c-ceil(size(W,2)/2))):min(end,ceil(c+size(W,2)/2)-1)),1,[]); W(i)=quantil(linw,msorted(ms==m(i)),1); end end end % 分位数计算函数 function Q=quantil(x, p, type) if (nargin<3), type=3; end; if (type~=1 && type~=3) error('Type must be 1 or 3'); end sz = size(x); idxp=find(p==1); % 如果 p 包含 1 if length(idxp)>0 idx = sub2ind(sz,[ones(size(idxp)),idxp]); Q(idx)=max(x(:)); end idxp=find(p==0); % 如果 p 包含 0 if length(idxp)>0 idx = sub2ind(sz,[ones(size(idxp)),idxp]); Q(idx)=min(x(:)); end idxp=find(p>0 & p<1); % 如果 p 在最大值和最小值之间 if length(idxp)>0 d = sort(x(:)); szx=length(d); midp1 = p(idxp)*(szx+type-1); midp2=cast(midp1,'int64')+1; midp1=cast(midp1,'int64'); idx1 = sub2ind(sz,[ones(size(idxp)),midp1]); idx2 = sub2ind(sz,[ones(size(idxp)),midp2]); Q(idx1)=d(midp1); Q(idx2)=d(midp2); delta=midp1-floor(midp1); idx = find(delta>0); % 丢失点是最常见的情况 if ~isempty(idx) idx=idx1(idx); Q(idx) = Q(idx).*(1-delta(idx)) + Q(idx2(idx)).*delta(idx); end end end % 中位数计算函数 function MED=median2(I) MED=quantil(I(:),0.50); end在运行该程序时,请确保以下事项:
- 将lena.jpg(或任何您希望在程序中使用的图像)保存在MATLAB当前工作目录中。
- 确保所有程序文件(包括此程序和以下程序代码中定义的函数)都被保存为.m文件,并在同一文件夹中。
此程序将加载lena图像并进行高斯模糊和 GDF降噪,最终输出降噪后的图像。
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.
Identical first prompt
- gpt-4-0314Compare side by side →
- gpt-4-0314Compare side by side →
- gpt-3.5-turbo-0301Compare side by side →
- gpt-3.5-turbo-0301Compare side by side →
- gpt-4-0314Compare side by side →
- gpt-4-0314Compare side by side →
- gpt-3.5-turbo-0301Compare side by side →
- gpt-4-0314Compare side by side →
- gpt-4-0314Compare side by side →
- gpt-4-0314Compare side by side →
- gpt-3.5-turbo-0301Compare side by side →
claude-instant-1Compare side by side →