how to detect corner of image and find coordinate of corner - I need M-file code

Please help , i had tried to find corner and coordinate of corner automatically,,, but still have problem to solve,, can u help me to make code-mfile,, i still confuse how to find corner and find coordinate automatically,, i have image tile square , how to detect 4 corner of tile ceramic and find value of coorninate 4 corner of tile ,

this link of image of tile : <http://tinypic.com/view.php?pic=2czer0o&s=6>

binary image of tile : <http://tinypic.com/view.php?pic=4q6gw4&s=6>

tile before croping and warping : <http://tinypic.com/view.php?pic=k4yyyd&s=6>

Answer 1

[Sean de Wolski]

How about:

doc cornermetric

as a place to start?

[KURNIAWAN]

how to detect edge corner? and find of coordinate

[david holguin]

where you find the document

[Image Analyst]

Try this page: Computer Vision Toolbox functions

Lots of corner-related functions there.

Answer 2

[matlab_image]

Please try with following links...... http://www.mathworks.com/matlabcentral/fileexchange/9272-harris-corner-detector/content/Harris.m http://www.mathworks.in/help/toolbox/images/ref/corner.html

[Walter Roberson]

Unfortunately we do not know if the poster is using a version of MATLAB new enough to have the "corner" routine. The poster was asked in one of the numerous threads the poster has started on this topic, but did not answer.

[KURNIAWAN]

i confuse with harris corner detector http://www.mathworks.com/matlabcentral/fileexchange/9272-harris-corner-detector/content/Harris.m

in m-file use imag.mat... how if i use image to process

Answer 3

[Image Analyst]

I thought from your prior messages on this tile that you needed the corners to know how to warp the image. Now it looks like you're saying that you warped it somehow without the corners. But unfortunately you warped it incorrectly so that it's still not square.

But if it were the binary image, you could simply use find() in the first and last row of the binary image to find the corners.

upperLeftCornerX = find(binaryImage(1,:), 1, 'first');
upperRightCornerX = find(binaryImage(1,:), 1, 'last');
lowerLeftCornerX = find(binaryImage(end,:), 1, 'first');
lowerRightCornerX = find(binaryImage(end,:), 1, 'last');

but again, you must already have the corners to do that incorrect warping so why not just do it correctly, using the code I gave you in the other posting? There is no need to correct the skewness only part way and then try to warp it again to get it all the way fixed.

[KURNIAWAN]

yah image analyst,, im still confuse,, can u give me into code like imtransform that u gived before,,, i just find one corner and know cordinate,, how about 4 corner n find pixel automatically,,

if u dont mind please teach me in code like u teach me to use imtransform

[Image Analyst]

I don't know what you need. Do you have the corner coordinates and just need to transform them into a rectangle? Or would you like me to do the whole thing for you (less likely to happen) where the corners are automatically found and then the the warping occurs. I could slap together a demo fairly quickly (later) where you click on the 4 corners with a mouse and then it warps it into a perfect rectangle. That is basically just taking the code I gave you and whipping it into a fully runnable script.

[KURNIAWAN]

thanks image analyst.. but i want to detect automatically without click on 4 corner and find the coordinat 4 corner,, i had tried but to hard to implement in code m-file,,

can u help me

[KURNIAWAN]

thats mean,, i want to detect the 4 corner automatically and find coordinat 4 corner,, then the coordinate will be used to know the distance of defect tile .. so i know the count value of defect on edge with the best tile reference

[KURNIAWAN]

this the problem : <http://tinypic.com/view.php?pic=9a9vdu&s=6>

so if the coordinate 4 corner automatically find, then i know defect of distance edge

[KURNIAWAN]

can u help me image analyst.. in whole code..

[Image Analyst]

No until sometime this weekend, but by then you'll probably have it all solved.