matlab数据集矩阵恢复图片,【求助】matlab怎样提取图片的数据呢，矩阵数据

function [clustCent,data2cluster,cluster2dataCell] = MeanShiftCluster(dataPts,bandWidth,plotFlag);

%perform MeanShift Clustering of data using a flat kernel

%

% ---INPUT---

% dataPts           - input data, (numDim x numPts)

% bandWidth         - is bandwidth parameter (scalar)

% plotFlag          - display output if 2 or 3 D    (logical)

% ---OUTPUT---

% clustCent         - is locations of cluster centers (numDim x numClust)

% data2cluster      - for every data point which cluster it belongs to (numPts)

% cluster2dataCell  - for every cluster which points are in it (numClust)

%

% Bryan Feldman 02/24/06

% MeanShift first appears in

% K. Funkunaga and L.D. Hosteler, "The Estimation of the Gradient of a

% Density Function, with Applications in Pattern Recognition"

%*** Check input ****

if nargin < 2

error('no bandwidth specified')

end

if nargin < 3

plotFlag = true;

plotFlag = false;

end

%**** Initialize stuff ***

[numDim,numPts] = size(dataPts);

numClust        = 0;

bandSq          = bandWidth^2;

initPtInds      = 1:numPts;

maxPos          = max(dataPts,[],2);                          %biggest size in each dimension

minPos          = min(dataPts,[],2);                          %smallest size in each dimension

boundBox        = maxPos-minPos;                        %bounding box size

sizeSpace       = norm(boundBox);                       %indicator of size of data space

stopThresh      = 1e-3*bandWidth;                       %when mean has converged

clustCent       = [];                                   %center of clust

beenVisitedFlag = zeros(1,numPts,'uint8');              %track if a points been seen already

numInitPts      = numPts;                               %number of points to posibaly use as initilization points

clusterVotes    = zeros(1,numPts,'uint16');             %used to resolve conflicts on cluster membership

while numInitPts

tempInd         = ceil( (numInitPts-1e-6)*rand);        %pick a random seed point

stInd           = initPtInds(tempInd);                  %use this point as start of mean

myMean          = dataPts(:,stInd);                           % intilize mean to this points location

myMembers       = [];                                   % points that will get added to this cluster

thisClusterVotes    = zeros(1,numPts,'uint16');         %used to resolve conflicts on cluster membership

while 1     %loop untill convergence

sqDistToAll = sum((repmat(myMean,1,numPts) - dataPts).^2);    %dist squared from mean to all points still active

inInds      = find(sqDistToAll < bandSq);               %points within bandWidth

thisClusterVotes(inInds) = thisClusterVotes(inInds)+1;  %add a vote for all the in points belonging to this cluster

myOldMean   = myMean;                                   %save the old mean

myMean      = mean(dataPts(:,inInds),2);                %compute the new mean

myMembers   = [myMembers inInds];                       %add any point within bandWidth to the cluster

beenVisitedFlag(myMembers) = 1;                         %mark that these points have been visited

%*** plot stuff ****

if plotFlag

figure(12345),clf,hold on

if numDim == 2

plot(dataPts(1,:),dataPts(2,:),'.')

plot(dataPts(1,myMembers),dataPts(2,myMembers),'ys')

plot(myMean(1),myMean(2),'go')

plot(myOldMean(1),myOldMean(2),'rd')

pause

end

end

%**** if mean doesn't move much stop this cluster ***

if norm(myMean-myOldMean) < stopThresh

%check for merge posibilities

mergeWith = 0;

for cN = 1:numClust

distToOther = norm(myMean-clustCent(:,cN));     %distance from posible new clust max to old clust max

if distToOther < bandWidth/2                    %if its within bandwidth/2 merge new and old

mergeWith = cN;

break;

end

end

if mergeWith > 0    % something to merge

clustCent(:,mergeWith)       = 0.5*(myMean+clustCent(:,mergeWith));             %record the max as the mean of the two merged (I know biased twoards new ones)

%clustMembsCell{mergeWith}    = unique([clustMembsCell{mergeWith} myMembers]);   %record which points inside

clusterVotes(mergeWith,:)    = clusterVotes(mergeWith,:) + thisClusterVotes;    %add these votes to the merged cluster

else    %its a new cluster

numClust                    = numClust+1;                   %increment clusters

clustCent(:,numClust)       = myMean;                       %record the mean

%clustMembsCell{numClust}    = myMembers;                    %store my members

clusterVotes(numClust,:)    = thisClusterVotes;

end

break;

end

end

initPtInds      = find(beenVisitedFlag == 0);           %we can initialize with any of the points not yet visited

numInitPts      = length(initPtInds);                   %number of active points in set

end

[val,data2cluster] = max(clusterVotes,[],1);                %a point belongs to the cluster with the most votes

%*** If they want the cluster2data cell find it for them

if nargout > 2

cluster2dataCell = cell(numClust,1);

for cN = 1:numClust

myMembers = find(data2cluster == cN);

cluster2dataCell{cN} = myMembers;

end

end