Vibe matlab

function [historyImages, historyBuffer] = vibeUpdate(buffer, shadowMap,updatingMask, historyImages, historyBuffer, param, ... jump, neighborX, neighborY, position) %% Parameters height = param.height; width = param.width; numberOfHistoryImages = param.numberOfHistoryImages; % numberOfSamples=param.numberOfSamples; % n=param.n; % step=param.step; % i=param.i; % t=param.t; % %% Update Model % for indY = 2:height - 1 % shift = floor(rand()*width) + 1; % indX = jump(shift) + 1; % while indX < width % value = buffer(indY, indX); % if updatingMask(indY, indX) == 0 %即为背景（黑色为0 白色为255） % % else %即为前景 % if(flags(indY,indX)==0) % counters(indY,indX)=counters(indY,indX)+1; % if(counters(indY,indX)>=ghosth)%则该点判定为持续 % sigmaB=var(historyBuffer(indY,indX)); % sigmaF=var % end % else % value=originalImage(indY,indX); % flags=flags*0;%%对新来一帧每隔像素点 % %%进行分析，如果改像素点为背景则直接使 % %%用新来一帧像素点进行更新，若该像素点 % %%为前景，且标识为1，则用保存的原始背景 % % 模型替换当前背景模型，且标识flag清零，存储器清空 % % end % end % if position(shift) <= numberOfHistoryImages % % historyImages{position(shift)}(indY, indX) = value; % historyImages{position(shift)}... % (indY + neighborY(shift), indX + neighborX(shift)) = value; % else % pos = position(shift) - numberOfHistoryImages; % historyBuffer{pos}(indY, indX) = value; % historyBuffer{pos}... % (indY + neighborY(shift), indX + neighborX(shift)) = value; % end % % shift = shift + 1; % indX = indX + jump(shift); % end % end %% Update Model for indY = 2:height - 1 shift = floor(rand()*width) + 1; indX = jump(shift) + 1; while indX < width if updatingMask(indY, indX) == 0 &&shadowMap(indY,indX)==1%即为背景（黑色为0 白色为255）且不是阴影 value = buffer(indY, indX); if position(shift) <= numberOfHistoryImages historyImages{position(shift)}(indY, indX) = value; historyImages{position(shift)}... (indY + neighborY(shift), indX + neighborX(shift)) = value; else pos = position(shift) - numberOfHistoryImages; historyBuffer{pos}(indY, indX) = value; historyBuffer{pos}... (indY + neighborY(shift), indX + neighborX(shift)) = value; end end shift = shift + 1; indX = indX + jump(shift); end end end % N=5; %考虑6帧的帧间差分法（需要读取前7帧） % start=5; %start=100，从第100+1帧开始连续读7帧 % threshold=50; % for k=1+start:N+1+start %处理从第101到第107帧 % avi(k).cdata=(avi(k).cdata); %将彩色图像转换为灰度图像 % %avi(k-start).cdata=avi(k).cdata; % end % [hang,lie]=size(avi(1+start).cdata); %以avi（1+start）.cdata的格式生成一个矩阵 % % alldiff=zeros(hang,lie,N); %生成一个三维的矩阵alldiff用于存储最终的各个帧的差分结果 % % for k=1+start:N+start % diff=abs(avi(k).cdata-avi(k+1).cdata); %邻帧差分 % % %idiff=diff>20; %二值化，阈值选择为20，阈值调整 % idiff=diff>threshold; %idiff中的数据位逻辑值，diff中的数值为unit8 % alldiff(:,:,k)=double(idiff); %存储各帧的差分结果，这里为什么要转换成double型的？？？？？ % end % end