哪位大侠可以帮帮忙把下面这段MATLAB的注释写下来哈，本人初学matlab，也可以发到邮箱哈xuxinanyin@163.com，感激不尽哈！（每句哈！）
%function [eigvector, eigvalue,Y] = KPCA(X,r,opts)
% Kernel Principal Component Analysis
% Written by Wei Yang
% [eigvector, eigvalue,Y] = KPCA(X,r,opts)
% Input:
% X: d*N data matrix;Each column vector of X is a sample vector.
% r: Dimensionality of reduced space (default: d)
% opts:   Struct value in Matlab. The fields in options that can be set:                 
%         KernelType  -  Choices are:
%                  'Gaussian'      - exp{-gamma(|x-y|^2)}
%                  'Polynomial'    - (x'*y)^d
%                  'PolyPlus'      - (x'*y+1)^d
%         gamma       -  parameter for Gaussian kernel
%         d           -  parameter for polynomial kernel
%
% Output:
% eigvector: N*r matrix;Each column is an embedding function, for a new
%            data point (column vector) x,  y = eigvector'*K(x,:)'
%            will be the embedding result of x.
%            K(x,:) = [K(x1,x),K(x2,x),...K(xN,x)]
% eigvalue: The sorted eigvalue of KPCA eigen-problem.
% Y       : Data matrix after the nonlinear transform
if nargin<1
  error('Not enough input arguments.')
end
[d,N]=size(X);
if nargin<2
  r=d;
end
%% Ensure r is not bigger than d
if r>d
    r=d;
end;
% Construct the Kernel matrix K
K =ConstructKernelMatrix(X,[],opts);
% Centering kernel matrix
One_N=ones(N)./N;
Kc = K - One_N*K - K*One_N + One_N*K*One_N;
clear One_N;
% Solve the eigenvalue problem N*lamda*alpha = K*alpha
if N>1000 && r
    % using eigs to speed up!
    opts.disp=0;
    [eigvector, eigvalue] = eigs(Kc,r,'la',opts);
    eigvalue = diag(eigvalue);
else
     [eigvector, eigvalue] = eig(Kc);
     eigvalue = diag(eigvalue);
     [junk, index] = sort(-eigvalue);
    eigvalue = eigvalue(index);
    eigvector = eigvector(:,index);
end
if r < length(eigvalue)
   eigvalue = eigvalue(1:r);
  eigvector = eigvector(:, 1:r);
end
% Only reserve the eigenvector with nonzero eigenvalues
maxEigValue = max(abs(eigvalue));
eigIdx = find(abs(eigvalue)/maxEigValue < 1e-6);
eigvalue (eigIdx) = [];
eigvector (:,eigIdx) = [];
% Normalizing eigenvector
for i=1:length(eigvalue)
    eigvector(:,i)=eigvector(:,i)/sqrt(eigvalue(i));
end;
if nargout >= 3
    % Projecting the data in lower dimensions
     Y = eigvector'*K;
end
%function K=ConstructKernelMatrix(X,Y,opts)
  % function K=ConstructKernelMatrix(X,Y,opts)
% Usage:
%   opts.KernelType='Gaussian';
% K = ConstructKernelMatrix(X,[],opts)
%   K = ConstructKernelMatrix(X,Y,opts)
%
% Written by Wei Yang
%
% Input:
% X: d*N data matrix;Each column vector of X is a sample vector.
% Y: d*M data matrix;Each column vector of Y is a sample vector.
% opts:   Struct value in Matlab. The fields in options that can be set:                 
%         KernelType  -  Choices are:
%                  'Gaussian'      - exp{-gamma(|x-y|^2)}
%                  'Polynomial'    - (x'*y)^d
%                  'PolyPlus'      - (x'*y+1)^d
%         gamma       -  parameter for Gaussian kernel
%         d           -  parameter for polynomial kernel
% Output:
% K N*N or N*M matrix
if nargin<1
   error('Not enough input arguments.')
end
if (~exist('opts','var'))
   opts = [];
else
  if ~isstruct(opts)
  error('parameter error!');
  end
end
N=size(X,2);
if isempty(Y)
  K=zeros(N,N);
else
   M=size(Y,2);
  if size(X,1)~=size(Y,1)
  error('Matrixes X and Y should have the same row dimensionality!');
end;
  K=zeros(N,M);
end;
%=================================================
if ~isfield(opts,'KernelType')
  opts.KernelType = 'Gaussian';   
end
switch lower(opts.KernelType)
case {lower('Gaussian')}        %  exp{-gamma(|x-y|^2)}
if ~isfield(opts,'gamma')
opts.gamma = 0.5;
   end
   case {lower('Polynomial')}      % (x'*y)^d
   if ~isfield(opts,'d')
opts.d = 1;
   end
case {lower('PolyPlus')}      % (x'*y+1)^d
    if ~isfield(opts,'d')
  opts.d = 1;
   end      
  otherwise
  error('KernelType does not exist!');
end
switch lower(opts.KernelType)
case {lower('Gaussian')}      
  if isempty(Y)
     for i=1:N
  for j=i:N
    dist = sum(((X(:,i) - X(:,j)).^2));
  temp=exp(-opts.gamma*dist);
  K(i,j)=temp;
if i~=j
   K(j,i)=temp;
  end;
end
  end
  else
for i=1:N
   for j=1:M
  dist = sum(((X(:,i) - Y(:,j)).^2));
  K(i,j)=exp(-opts.gamma*dist);                  
end
end
end      
  case {lower('Polynomial')}     
  if isempty(Y)
for i=1:N
for j=i:N                    
temp=(X(:,i)'*X(:,j))^opts.d;
  K(i,j)=temp;
   if i~=j
  K(j,i)=temp;
  end;
  end
end
  else
   for i=1:N
  for j=1:M                                       
  K(i,j)=(X(:,i)'*Y(:,j))^opts.d;
  end
end
  end      
  case {lower('PolyPlus')}     
if isempty(Y)
for i=1:N
  for j=i:N                    
  temp=(X(:,i)'*X(:,j)+1)^opts.d;
K(i,j)=temp;
   if i~=j
  K(j,i)=temp;
end;
end
  else
   else
  for i=1:N
  for j=1:M                                       
   K(i,j)=(X(:,i)'*Y(:,j)+1)^opts.d;
  end
  end
   end      
   otherwise
   error('KernelType does not exist!');
end