Error using ==> corr at 103 X and Y must have the same number of rows. Error in ==> motion_corr2 at 108 [a, F] = corr(p1, p2, C, 'sdthresh', sdthresh, corr_opts{:}); Error in ==> symmetric_match at 3 [p2a,p1a,a2,Fa] = motion_cor​r2(f2,k2,f​1,k1,im2,

Error using ==> corr at 103 X and Y must have the same number of rows.
Error in ==> motion_corr2 at 108 [a, F] = corr(p1, p2, C, 'sdthresh', sdthresh, corr_opts{:});
Error in ==> symmetric_match at 3 [p2a,p1a,a2,Fa] = motion_corr2(f2,k2,f1,k1,im2,

function corr.m

if (nargin < 2) || ischar(varargin{1})
    corrXX = true;
    y = x;
    p2 = p1;

% Both x and y given, compute the pairwise rank cross correlations
else
    y = varargin{1};
    varargin = varargin(2:end);
    if size(y,1) ~= n
        error('stats:corr:InputSizeMismatch', ...
              'X and Y must have the same number of rows.');
    end
    corrXX = false;
    p2 = size(y,2);
end
outClass = superiorfloat(x,y);

function motion_corr2.m

function [p1, p2 , a, F] = motion_corr2(f1,k1,f2,k2,im1,im2, varargin)

[p1, ...
 smoothing, ...
 nmsrad, ...
 rthresh, ...
 rthresh2, ...
 sdthresh, ...
 dthresh, ...
 corr_opts] = process_options(varargin, 'p1', [], ...
                                        'smoothing', 2, ...
                                        'nmsrad', 2, ...
                                        'rthresh', 0.3, ...
                                        'rthresh2', nan, ...
                      'sdthresh', 1e-2, ...
                      'dthresh', 30);

if (isnan(rthresh2)) rthresh2 = rthresh / 2.0; end

C = make_cost(k1,k2);

p1 = f1(:,1:2); %create homogeneous coordinates 
p2 = f2(:,1:2);
p1(:,3) = 1;
p2(:,3) = 1;

[a, F] = corr(p1, p2, C, 'sdthresh', sdthresh, corr_opts{:});

symmetric_match.m

[p2a,p1a,a2,Fa] = motion_corr2(f2,k2,f1,k1,im2,im1, 'sdthresh', 1e-4);
'done with #1';
[p1b,p2b,a1,Fb] = motion_corr2(f1,k1,f2,k2,im1,im2,'sdthresh', 1e-4);

r=zeros(size(a1,1),1);
for i=1:size(a1,1)
    if a1(i)>0
        if a2(a1(i)) == i
            r(i)=1;
        end
    end
end

showfeatures(f2,im2);

showfeatures(f1,im1);
hold on
for i=1:size(p1b,1)
    x = p1b(i,1);
    y = p1b(i,2);
    if a1(i)~=0 & r(i)>0
        u = p2b(a1(i),1)-p1b(i,1);
        v = p2b(a1(i),2)-p1b(i,2);
        plot([x x+u],[y y+v],'y');
    end
end

pt1 = p1b(find(r),:);
pt2 = p2b(a1(find(r)),:);
w1 = f1(find(r),4);
w2 = f2(a1(find(r)),4);

g=sampson(pt1, pt2, Fb);
m = mean(g);

fk1 = k1(find(r),:);
fk2 = k2(a1(find(r)),:);

for i=1:size(fk1,1)
    c(i) = sum((fk1(i,:)-fk2(i,:)).^2);
end

cavg = mean(c);

pt1 = pt1(:,1:2)';
pt2 = pt2(:,1:2)';
pt1 = (pt1-1)/2;
pt2 = (pt2-1)/2;
w1 = (w1/2)';
w2 = (w2/2)';

[Walter Roberson]

Please read the guide to tags and add good tags to this question. "confuse" does not describe anything about the topic of your question. See http://www.mathworks.co.uk/matlabcentral/answers/43073-a-guide-to-tags

[Walter Roberson]

Do not open new questions for the same topic. Instead, reply with the updated information.

Answer 1

[Walter Roberson]

In your routine symmetric_match, the first thing you do is call motion_corr2, passing in a number of parameters. The first and third parameters, f1 and f2, need to have the same number of rows but they do not.

Is symmetric_match a function or a script? Did you forget to paste the "function" line from the start of the routine? How are you running symmetric_match ?

[Luo Zhangquan]

How can i change f1 and f2, to make them have the same number of rows

[Walter Roberson]

What are f1 and f2 at that point? How are you executing symmetric_match ? The code you show us does not define f1 or f2.

[Luo Zhangquan]

symmetric_match is a script.

imgo1 = imread('A1.jpg');
img1 = imresize(imgo1,2,'bilinear');
imgo2 = imread('B2.jpg');
img2 = imresize(imgo2, 2, 'bilinear');
[features,pyr,imp,keys] = detect_features(img1,1.5,0,3,4,4,4,.04,5);
[features2,pyr2,imp2,keys2] = detect_features(img2,1.5,0,3,4,4,4,.04,5);
kill_edges
im1 = rgb2gray(img1);
im2 = rgb2gray(img2);
symmetric_match
m
cavg
size(fk1)

kill_edges.m

f1 = features(find(features(:,5)>0),:);
f2 = features2(find(features2(:,5)>0),:);

k1 = keys(find(features(:,5)>0),:);
k2 = keys2(find(features2(:,5)>0),:);

[Walter Roberson]

Notice that you have the line

symmetric_match

as part of your file symmetric_match.m . When your program gets to that point in the file symmetric_match.m, it would run symmetric_match again, including getting to the line symmetric_match, which would cause it to run the file again... and so on, never ending.

[Walter Roberson]

Why do you think that detect_features is going to find exactly the same number of positive features for img1 and it would for img2 ? If the number of detected features is different by even 1, then you are going to be trying to do motion correlation between different number of features, and that is going to cause the problems you are getting with corr()

[Luo Zhangquan]

symmetric_match is a script symmetric_match.m

[p2a,p1a,a2,Fa] = motion_corr2(f2,k2,f1,k1,im2,im1, 'sdthresh', 1e-4); 'done with #1'; [p1b,p2b,a1,Fb] = motion_corr2(f1,k1,f2,k2,im1,im2,'sdthresh', 1e-4); r=zeros(size(a1,1),1); for i=1:size(a1,1) if a1(i)>0 if a2(a1(i)) == i r(i)=1; end end end showfeatures(f2,im2); showfeatures(f1,im1); hold on for i=1:size(p1b,1) x = p1b(i,1); y = p1b(i,2); if a1(i)~=0 & r(i)>0 u = p2b(a1(i),1)-p1b(i,1); v = p2b(a1(i),2)-p1b(i,2); plot([x x+u],[y y+v],'y'); end end pt1 = p1b(find(r),:); pt2 = p2b(a1(find(r)),:); w1 = f1(find(r),4); w2 = f2(a1(find(r)),4); g=sampson(pt1, pt2, Fb); m = mean(g); fk1 = k1(find(r),:); fk2 = k2(a1(find(r)),:); for i=1:size(fk1,1) c(i) = sum((fk1(i,:)-fk2(i,:)).^2); end cavg = mean(c); pt1 = pt1(:,1:2)'; pt2 = pt2(:,1:2)'; pt1 = (pt1-1)/2; pt2 = (pt2-1)/2; w1 = (w1/2)'; w2 = (w2/2)';

function motion_corr2.m

function [p1, p2 , a, F] = motion_corr2(f1,k1,f2,k2,im1,im2, varargin) [p1, ... smoothing, ... nmsrad, ... rthresh, ... rthresh2, ... sdthresh, ... dthresh, ... corr_opts] = process_options(varargin, 'p1', [], ... 'smoothing', 2, ... 'nmsrad', 2, ... 'rthresh', 0.3, ... 'rthresh2', nan, ... 'sdthresh', 1e-2, ... 'dthresh', 30); if (isnan(rthresh2)) rthresh2 = rthresh / 2.0; end C = make_cost(k1,k2); p1 = f1(:,1:2); %create homogeneous coordinates p2 = f2(:,1:2); p1(:,3) = 1; p2(:,3) = 1; [a, F] = corr(p1, p2, C, 'sdthresh', sdthresh, corr_opts{:});

I ues the following codes to call the script symmetric_match.m and the function corr.m is included in the file toolbox starts imgo1 = imread('A1.jpg'); img1 = imresize(imgo1,2,'bilinear'); imgo2 = imread('B2.jpg'); img2 = imresize(imgo2, 2, 'bilinear'); [features,pyr,imp,keys] = detect_features(img1,1.5,0,3,4,4,4,.04,5); [features2,pyr2,imp2,keys2] = detect_features(img2,1.5,0,3,4,4,4,.04,5); kill_edges im1 = rgb2gray(img1); im2 = rgb2gray(img2); symmetric_match m cavg size(fk1)

and the matlab is getting error:??? Error using ==> corr at 103 X and Y must have the same number of rows.

Error in ==> motion_corr2 at 108 [a, F] = corr(p1, p2, C, 'sdthresh', sdthresh, corr_opts{:});

Error in ==> symmetric_match at 3 [p2a,p1a,a2,Fa] = motion_corr2(f2,k2,f1,k1,im2,im1, 'sdthresh', 1e-4);

[Walter Roberson]

http://www.mathworks.com/matlabcentral/answers/13205-tutorial-how-to-format-your-question-with-markup

[Walter Roberson]

Why do you think that detect_features is going to find exactly the same number of positive features for img1 and it would for img2 ? If the number of detected features is different by even 1, then you are going to be trying to do motion correlation between different number of features, and that is going to cause the problems you are getting with corr()