i have a binary image with gird i want to select each segments in a grid

Help me to extract or select a region in a grid because i want to find foreground pixels in each grid in the attached image.

[Walter Roberson]

In your sample image, if you look at the top horizontal line, on the right hand side of the right-most vertical line, the horizontal line appears to curve a bit instead of continuing straight. Is that correct?

It would be much easier to break the image into segments if the horizontal and vertical grid lines continued continued completely straight. It also avoids the question of whether the little dip in the top right rectangle near that intersection should be counted as belonging to the top right rectangle or should instead be counted as belonging to the rectangle below it as would be expected by straight lines.

[Raghu Rao]

I want to find foreground pixel in each grid

[Image Analyst]

They're already "found" - they're white aren't they? Exactly what do you want? Their gray level? It's true or 1 or 255 for all of them. Their coordinates? Why - it's not needed?

[Raghu Rao]

I want to find the centre of mass of each segment in a grid because I want to add the nodes based on centre of mass

[Raghu Rao]

Please go through the above attached image, am working on hand written word image and I want to extract graph from it using grid strategy.

[Walter Roberson]

We need an answer to the question about whether the grid squares should be defined by the not-quite-straight lines or if they should be defined by interpolating straight lines.

https://www.mathworks.com/matlabcentral/answers/325876-i-have-a-binary-image-with-gird-i-want-to-select-each-segments-in-a-grid#comment_430133

[Image Analyst]

What are the rows and columns of your grid lines? You know how to compute the center of mass - it's just the mean of the rows and columns in the grid.

thisGridTile = binaryImage(row1:row2, column1:column2);
[y, x] = find(thisGridTile);
centroidx = mean(x);
centroidy = mean(y);

[Raghu Rao]

My question is I want to divide a binary image into CxR equal size and in each segment Sij I want to to find foreground

Answer 1

[Image Analyst]

I don't know exactly what that means, but to extract a grid tile from the large image, if you know the rows and columns, you can simply do this:

thisGridTile = binaryImage(row1:row2, column1:column2);

beyond that then you will need to specify exactly what information you want when you say you want to "find" or "select" the white foreground pixels in each tile. In my mind they're already automatically selected. What else more do you want???

[Raghu Rao]

Sorry, am working on handwritten word image in grid. When I select an image I want to convert that image into binary after converting the image I want to add rows and columns like C = width of binary image / w, where w is user defined width and R = height of binary image / h, where h is user defined height, and finding C And R values I want to divide an binary image of equal size CxR, after dividing in each segment Sij for i:C for j:R In each segment I want to find foreground pixels based on foreground pixels I want to find the centre of mass after finding the centre of mass I want to add node to where centre of present in the segment Sij

Answer 2

[Raghu Rao]

Please go through the above attached image, am working on hand written word image and I want to extract graph from it using grid strategy.

Answer 3

[Walter Roberson]

blockproc() large blocks over the image. Have blockproc call regionprops on the binary negation of the image to determine the centroid of the black pixels.