how can i find center line of weld defect image, and how recoginze object

 In welding  defect image found 3 objects  want center line  these  objects , and may i ask  how  can i  recognize  these objects  for   welding  defect imge  ,i want example code  for  recognize object and find center line of those .

age ,

Answer 1

[Image Analyst]

You say you found them already, and I can tell this is the case because you must have used bwboundaries on the binary image so that you could plot the outlines in red. So just pass that binary image into regionprops and ask for the centroid. I'm not sure how you define centerline, but you can also ask for orientation (the angle) and the majoraxis length. From those 3 measurements you can probably get "the center line" according to your definition.

See Steve Eddin's Image Processing blog: http://blogs.mathworks.com/steve/2010/07/30/visualizing-regionprops-ellipse-measurements/ for an example.

[thet aung]

thank for reply, i want centerline of 3 binary image ,coz for classifine object , and how can i recognize founded 3 object in weld seam have defect or non defect ,(exp as neural network ,Som,etc,) i need object classification for result code

clc
clear all
close all
%%load image 
[filename, pathname]=uigetfile({'*.*'},'Pick an Image file');
I=imread([pathname,filename]);
imshow(I);title(' Original image ');
%%change gray image
I2= rgb2gray(I);
imshow (I2),title('gray image ');
%%ENCHANT
I_inden=imadjust(I2,[0.3 0.7],[]);
imshow(I_inden)
%%FILTER
I_med=medfilt2(I_inden,[5 5]);
%%BINARY
I_bw=im2bw(I_med,0.15);
imshow(I_bw);
%%MOPHOLOGICAL
se=strel('disk',1);
I_dilate=imclose(I_bw,se);figure,imshow(I_dilate)
se=strel('disk',1);
I_erode=imdilate(I_dilate,se);
figure,imshow(I_erode),title('morphological erode image');
im_fill=imfill(I_erode,'holes');
border_clear=imclearborder(im_fill,8);
clear_border=bwareaopen(border_clear,300);
imshow(clear_border)
[B L]=bwboundaries(clear_border,'noholes');
[L Ne]=bwlabel(clear_border);
propied=regionprops(L);
numberOfdefect = size(propied, 1);
coloredLabels2 = label2rgb (L, 'hsv', 'black', 'shuffle');
imshow(coloredLabels2)

imshow(I)
hold on
for n=1:size(propied,1)
    rectangle('Position',propied(n).BoundingBox,'EdgeColor','r','LineWidth',2);
end
[imx,imy]=size(L);
for i=1:imx
   srows(i)=sum(L(i,:));
end
figure, plot (srows);
[imx,imy]=size(L);
for i=1:imy
   scolumns(i)=sum(L(:,i));
end
figure,subplot(2,1,2),plot (scolumns);
grid on;
hold on;
title('vertical defect image projection')
subplot(2,1,1)
imagesc(I)
hold on; contour(clear_border,[0 0],'r','linewidth',2); title('Image segment  final');hold off;

[Image Analyst]

People can run your code on your image only if you upload one of your images (not a screenshot).

[thet aung]

thank for reply , this is original image weld seam

[thet aung]

this is also weld seam image 2

[thet aung]

[ah gh]

Hi Dear Thet Aung Your sample code in detection of welding defects is very good and helpful for me. Do you write any classifier for this weld detection?? Thanks