图像几何变换.rar

function rI = myImrotate(I,angle,varargin) %获取必要的计算参数 [method,bbox] = ParseInputs(varargin{:}); %角度转换为弧度 angle = angle * pi / 180; %由结果影像范围，计算几何转换关系 R = CalculateTransformMatrix(angle); %R为旋转矩阵 %计算结果影像 inSz = size(I); %输入图像的尺寸 if strcmpi(bbox,'loose') %由四个角点变换后的位置，计算结果影像的大小 xIn = [1;1;inSz(2);inSz(2)]; yIn = [1;inSz(1);inSz(1);1]; [xOut,yOut] = In2Out(R,xIn,yIn); %像素取整 xOutMin = floor( min(xOut) ); %取不大于它的最小整数 yOutMin = floor( min(yOut) ); xOutMax = ceil( max(xOut) ); yOutMax = ceil( max(yOut) ); %取不小于它的最小整数 outSz = [yOutMax, xOutMax] - [yOutMin, xOutMin] + 1; %记录结果影像的左上角点，相对于原始影像坐标系的位置 xOffset = xOutMin; yOffset = yOutMin; elseif strcmpi(bbox,'crop') outSz = inSz(1:2); xOffset = 1; yOffset = 1; end %原始影像数据类型 dataType = class(I); %波段数 bandNum = size(I,3); rI = zeros([outSz,bandNum],dataType); %计算结果影像的格网点位置 [xOut,yOut] = meshgrid(1:outSz(2),1:outSz(1)); %计算结果影像格网点，相对于原始影像坐标系的位置 xOut = xOut(:) - 1 + xOffset; yOut = yOut(:) - 1 + yOffset; [xIn,yIn] = Out2In(R,xOut,yOut); %对应的原始影像坐标 for k = 1:size(I,3)%按波段遍历 %改变数据类型，满足函数interp2要求，构造临时变量tmp tmp = double( I(:,:,k) ); %采用method方法，内插灰度 tmp = interp2(tmp, xIn, yIn, method); rI(:,:,k) = reshape(tmp, outSz); end end function [method,bbox] = ParseInputs(varargin) inputNum = length(varargin);%输入参数的个数 method = 'nearest'; bbox = 'loose'; switch inputNum case 0 case 1 method = varargin{1}; case 2 method = varargin{1}; bbox = varargin{2}; otherwise error('Too many input arguments.'); end end function R = CalculateTransformMatrix(angle) R = [ cos(angle), -sin(angle); sin(angle), cos(angle); ]; end function [xout,yout] = In2Out(R,xin,yin) %这里默认输入的x1,y1分别是1列向量，即n-by-1的矩阵 p = [xin,yin] * R; xout = p(:,1); yout = p(:,2); end function [xin,yin] = Out2In(R,xout,yout) p = [xout,yout] / R; xin = p(:,1); yin = p(:,2); end