why X and Y coordinates must have the same size ?

hello,

I have an image and a shape and I need to clip the image based on the coordinates of the shape but when I do this I am having an error says: x and y must be the same size how can I make them equal? or any idea of solving this issue please ?

Thanks

[KSSV]

Why don't you provide your code.....so that we can have more clear idea..

[Reema Alhassan]

the image is very large I can't attach it here I have attached the error that I'm having and here is the code :

info = geotiffinfo('SVDNB_npp_20121201-20121231_75N060W_vcmcfg_v10_c201601041440.avg_rade9h.tif');
%disp(info);
[x1,y1]= pixcenters(info);
%pix2map(info.RefMatrix,1,1)
shape = shaperead(fullfile('KTAF_sat_shapefiles','shapefiles','SAU0.csv.shp'));
x2=shape.X;
y2=shape.Y;
%numel(x)
%numel(y)
[x3,y3]= polyclip(x1,y1,x2,y2,1);
function [X,Y] = polyclip(x1,y1,x2,y2,method) % POLYCLIP  Clip two polygons (intersect, union, xor, diff).
%
%  [X,Y] = POLYCLIP(X1,Y1,X2,Y2,METHOD) clips the two polygons given by the %  points (X1(i),Y1(i)) and (X2(j),Y2(j)) according to the METHOD:
%         METHOD = 0 or 'dif' ==> difference (P1-P2)
%         METHOD = 1 or 'int' ==> intersection
%         METHOD = 2 or 'xor' ==> Xor
%         METHOD = 3 or 'uni' ==> Union
%  X and Y are cell arrays (because the result may be multiple polygons) %  containing the x and y coordinates of the resulting polygon(s).
%
%  XY = POLYCLIP([X1 Y1],[X2 Y2],METHOD) does the same assuming the Xi and Yi %  are column vectors.  The result is {X Y}.
%
% See also CLIPPER, POLYOUT.
% Copyright (c)2015-17, Prof. Erik A. Johnson <[e-mail]>, 01/29/17
% 09/13/15  EAJ  Initial code
% 01/28/17  EAJ  Update for newer MATLAB versions % 01/29/17  EAJ  Fix: test for METHOD as string, single output
narginchk(2,5);
if any(nargin==[2 4]), method=[]; end
if any(nargin==[2 3]) % POLYCLIP([X1 Y1],[X2 Y2],METHOD)
    if size(x1,2)~=2 || size(y1,2)~=2 || ~isnumeric(x1) || ~isnumeric(y1),
        error('The [X1 Y1] and [X2 Y2] matrices must be numeric two-column matrices.');
    end
    if nargin == 3
        method = x2;
    else
        method = [];
    end
    y2=y1(:,2); x2=y1(:,1);
    y1=x1(:,2); x1=x1(:,1);
else
    if ~isnumeric(x1) || ~isnumeric(y1) || ~isnumeric(x2) || ~isnumeric(y2)
        error('The polygon coordinates must be numeric matrices.')
    elseif numel(x1)~=length(x1) || numel(y1)~=length(y1) || numel(x2)~=length(x2) || numel(y2)~=length(y2)
        error('Function only handles the clipping of one polygon with another, not multiple polygons with multiple polygons');
    elseif numel(x1)~=numel(y1) || numel(x2)~=numel(y2)
        error('X1 must be the same size as Y1, and X2 the same size as Y2.');
    end;
    if length(x1)==numel(x1), x1=x1(:); y1=y1(:); end; % ensure column
    if length(x2)==numel(x2), x2=x2(:); y2=y2(:); end; % ensure column
    if nargin == 4
        method = [];
   end
end;
if isempty(method), method=0; end
if isstr(method)
    find(lower(method(1))==['dixu']) - 1;
    if isempty(ans)
        error('METHOD must be an integer (0 to 3) or {''diff'',''int'',''xor'',''union''}.');
    end;
    method = ans;
end
if ~isscalar(method) || ~any(method==0:3)
    error('METHOD must be 0, 1, 2 or 3.')
end;
scale = 2^32;
% leave this in array format just in case we later adapt this
pack   = @(p) arrayfun(@(x) struct('x', int64(x.x*scale),'y', int64(x.y*scale)),p);
unpack = @(p) arrayfun(@(x) struct('x',double(x.x)/scale,'y',double(x.y)/scale),p);
poly1 = struct('x',num2cell(x1,1),'y',num2cell(y1,1));
poly2 = struct('x',num2cell(x2,1),'y',num2cell(y2,1));
assert(numel(poly1)==1, 'Only single polygon clippings are allowed in the current code');
poly3 = unpack(clipper(pack(poly1),pack(poly2),method));
if isempty(poly3)
    x = cell(1,0);
    y = cell(1,0);
else
    x = {poly3.x};
    y = {poly3.y};
end
% % package into single polygon with NaN's separating individual polygons % lens = cellfun(@numel,{poly3.x}'); % cellfun(@(x,l) [x;NaN(max(lens)-l,1)], {poly3.x}', num2cell(lens), 'UniformOutput',false); x=cat(2,ans{:}); % cellfun(@(y,l) [y;NaN(max(lens)-l,1)], {poly3.y}', num2cell(lens), 'UniformOutput',false); y=cat(2,ans{:});
if nargout>=2
    X = x;
    Y = y;
else
    % X = cellfun(@(x,y) [x(:) y(:)],x,y,'UniformOutput',false);
    X = {x y};
end
end

[Walter Roberson]

Is the actual error that you are getting

X1 must be the same size as Y1, and X2 the same size as Y2.

or are you getting a different error ?

[Reema Alhassan]

yes,this is the error

[Walter Roberson]

What shows up for

size(x1)
size(y1)
size(x2)
size(y2)

Answer 1

[Walter Roberson]

I suggest using poly2mask() to create mask (binary image) that you can then multiply by the original image to get the clipped image. You might always want find the bounding box of the mask and use that to crop the result.

Your x and y coordinates will need to be the same length for poly2mask.

That might involve deliberately repeating some coordinates. For example to express y ranging from 10 to 20 at x = 7, coming from (0,0), then you cannot just code x = [0 7], y = [0 10 20] -- each y must be matched with an x. You would instead code x = [0 7 7], y = [0 10 20]

[Reema Alhassan]

yes I got it, but the size of the coordinates very large I can't do it manually also I have to deal with more than one image .. is this something related to projection ? because I'm using geotiff image (satellite image ) and when I check the sizes of the x and y coordinate I always get different sizes ..

[Reema Alhassan]

hey

here when you said "then multiply by the original image to get the clipped image" do you mean multiply it with the coordinates?

thanks

[Image Analyst]

Try

mask = poly2mask(x, y, rows, columns);
% Mask the image using bsxfun() function to multiply the mask by each channel individually.
maskedRgbImage = bsxfun(@times, rgbImage, cast(mask, 'like', rgbImage));

[Reema Alhassan]

what do you mean here by rows and columns ? could you give me an example please?

[Image Analyst]

You said "I have an image" and so rows and columns would be the number of rows and columns in your image.

[Reema Alhassan]

this is what I get :(

[Reema Alhassan]

actually I have tried this code and it is working for the small images :

 R = geotiffinfo('clipped2.tif');
f= geoimread('clipped2.tif');
% Get x,y locations of pixels:
[x,y] = pixcenters(R);
% Convert x,y arrays to grid:
[X,Y] = meshgrid(x,y);
roi = shaperead(fullfile('KTAF_sat_shapefiles','shapefiles','SAU0.csv.shp'));
% Remove trailing nan from shapefile
rx = roi.X(1:end-1);
ry = roi.Y(1:end-1);
mask =inpolygon(X,Y,rx,ry);
t= mask.*f;
sumOfValues= sum(t(:));
sumOfValues1=sprintf('%f', sumOfValues)

but I have to deal with larger images about 4GB so when I run it with large image the computer is crashing in either meshgrid() or in inpolygon()

[Walter Roberson]

Is it possible that each of your shapes has several embedded NaN? It is not uncommon to pack several shapes into one vector, especially if there are discontinuous pieces (such as if lakes or islands are outlined as well.)

[Reema Alhassan]

I can take the coordinates of each polygon and deal with each one alone so this is not a problem

[Reema Alhassan]

my problem is that the geotiff image is very large so I need to clip it first based on the boundingBox of the shape to make it smaller and then apply the function inpolygon() but I couldn't find a way to clip it and make a new geotiff image if you know how to do this could you help me please?

Thanks

[Walter Roberson]

"Expecting input number 1, X, to be finite"

is an error you are getting because you are feeding in a polygon that has NaN inside it, to mark the end of a segment such as to move over to draw a lake or island. You need to break up your inputs to poly2mask, creating several masks and or'ing them together.