I have an image with yellow background containing a random figure as shown in figure:
The random figure is divided by black lines into image pieces. Now each piece can be represented separately as a square containing that piece image with transparent background.
My question if it possible to find the coordinates of each piece algorithmically in the original image?
I am writing this application in Java.
I don't have much idea about the graphics. If its possible then please elaborate little bit.
Presuming the images look mostly like what you have here
Loop
Find a Red pixel
If found
flood fill red to non-red at this point, remembering region
create output image from this region
else
You are done
Use connected component labeling on the binary image (threshold your current image).
I used MATLAB to threshold the image, and run a labeling algorithm. Then I used region properties to find the centroid of each connected component (which are the image pieces you need). The following is the labeled image with the black stars representing the centroid of each piece:
Related
I am currently working on a solution for an automatic image edit.
And I have used Canny Edge Detection and Closing.
But What I ultimately want to accomplish is to find all the rectangles from blueprint and fill them with an image what I have.
I want to know the process that I need to take, not the exact code or solution!
Please suggest me what steps should I take to accomplish it, thx.
Image that has the rectangles
Image that needs to go into all the rectangles
what i have done so far
2018-02-12 EDITTED(clean rectangle detected)
// I have done finding the rectangles and drawing lines over them, but the result is not really reliable than I expected(it draws line on rectangles those are not a parking space), and I do not know how to put an image on those rectangle instead of drawing line on them. please help me out!
P.S : Only in JAVA please !
In your case, you don't need Canny. Your image is really clean and edges are really visible already. A simple Threshold, will work better.
For rectangle detection take a look at this example (included with opencv) that uses findcontours and checks the angles: https://github.com/opencv/opencv/blob/master/samples/cpp/squares.cpp
In your case, you can skip the Canny step because you don't have gradients. You may need to modify the filtering, like side dimensions for your case.
Once the rectangles look good, you just need to copy your image in the location. If rectangles are rotated, you will have to rotate your image as well.
EDIT:
To copy the small image onto the big image you can do the following
Mat submatImg = bigImage.submat(new Rect(x, y, smallImage.width(), smallImage.height());
smallImage.copyTo(submatImg);
If you need to do resizing and rotations, take a look at geometric transformations
I have the following problem:
I have written a plugin for ImageJ, which sets specific pixels (the blue ones = cells) to the color red.
Result after setting the blue pixels red
Now I also want to set also all the pixels, which are inside a marked area, to red.
So I don't want to have any holes.
Example for a hole:
I searched the web for hours, but I didn't find anything. I hope somebody knows how I should deal with this issue.
Perhaps it is because the value of blue pixels varies. A better way would be to detect edges and then use flood fill algorithm to fill in the regions inside the boundary.
Found one sample code here http://imagej.net/Level_Sets with github source https://github.com/fiji/level_sets/.
I am using a Java application to display an image on the screen. I also am using an eye-tracker device which records the absolute pixel X,Y locations where the person is looking on the screen.
However, what I need to do is convert these X,Y coordinates from the screen positions into the X,Y locations of the image. In other words, somehow I need to figure out that (just an example) 482, 458 translates to pixel 1,1 (the upper left pixel) of the image.
How can I determine the image's placement on the screen (not relative to anything)?
I saw a few posts about "getComponentLocation" and some other APIs, but in my experimentation with these, they seem to be giving coordinates relative to the window. I have also had problems with that because the 1,1 coordinate that they give is within the window, and there is actually a bar at the top of the window (that has the title and the close and minimize buttons) whose width I do not know, so I cannot easily translate.
Surely there must be a way to get the absolute pixel location on the screen of a component?
If we are talking about Swing/AWT application than class java.awt.Component has method getLocationOnScreen which seemed to do what you want
And yes as #RealSkeptic mentioned in comments to question:
SwingUtilities.html#convertPointFromScreen
will do all this work for you considering components hierarchy
From the above image if I want a portion behind the RED Rectangle I can easily get it,
but the issue I cannot get the portion behind the Yellow Rectangle because it is rotated.
So how can I get a portion of an image from a rotated shape on it?
For example my goal is to get a portion of an Image where the rectangle is located on the image. if someone rotates this rectangle by an x degree [in whatever direction] then it is getting difficult to extract the exact portion of an image after applying rotation.
Any suggestions?
Here a more lengthy description of a possible approach. I do not know the Java2D drawing API very well but if I remember correctly it has the capabilities to do what is required.
First you have to figure out the translation and rotation of the subregion you want compared to an equally size rectangle located straight in the upper left corner in the image. Then invert this transformation.
Make a graphics context which is backed by a bitmap in memory. This one should have the size of the subimage you want. Setup the inverse transformation you calculated earlier on the context and draw your image at position 0,0. As Java2D will take the transformation into account you should now get the sub image you want in the memory bitmap.
Mihir, I think you might be getting distracted by the rotation/AffineTransform aspects of this challenge and it is leading you down the wrong road. Also keep in mind that I don't totally know what you mean by "get" here -- do you want to save out the highlighted region to an image? Do you want to render it as a watermark on another image? etc... I'll just try and answer in the general case to get you down the right track.
What you want is the content from the image defined by the polygon in yellow in your image above; ignoring the fact that it looks like a rotated rectangle.
It is late and I am missing a step in here, but I think this will get you 90% of the way there and clarify the last piece (Graphics2D.setClip) that you need.
Create a java.awt.Polygon that defines the region around the area you want.
Use getBounds() or getBounds2D() to get the width/height of the bounding box required to hold this Polygon when rendered out into a rectangle. (e.g. boundingBox)
Create a new BufferedImage with these width/height values.
Get the Graphics2D from the new BufferedImage (e.g. newG2)
newG2.drawImage(originalImage, boundingBox.x, boundingBox.y, boundingBox.width, boundingBox.height)
NOTE This is where my memory is failing me; at some point you need to set the clip on newG2 (newG2.setClip(someShape)) so when the bounding box is rendered into it, you don't get the full bounding box of graphics rendered in, but instead some subset as defined by the yellow outline.
One easy way to do this is to create two Polygon's:
poly1 = a java.awt.Polygon that defines the yellow selection in the ORIGINAL image.
poly2 = a java.awt.Polygon that defines the exact same shape of Polygon, but shifted to a 0,0 origin point.
poly1 is used to get the bounding box to copy out the full bounding box that encompasses the content selected in yellow (and extra content around it)
poly2 is used to set the clip on the target Graphics2D (newG2) so when the bounding box is rendered into it, we clip back out everything outside of that Yellow shape so we just get the content in Yellow. You'll likely want to use an ARGB image type and set the background of the target image as transparent otherwise you'll get a black fill color.
I think this is the right direction for the clips; I was up to my eyeballs in Java2D for years and years but have been out of it for a while and forget if this will give you exactly what you want or not; you might need to tweak it around, but these are all the tools you need.
Here's an image:
I would like to know how i can set the black circle to white and the rest to black.
(So segment the black circle within the white area).
I know i can invert the image and the circle will be white ... but so will the entire black part that is seen in this image.
If i would have to do this in matlab i would do a connected component operation and check the circularity of the BLOBs. Though I have to do this in opencv (javacv to be precise.)
Is there an easy way of doing this in opencv (javacv).
Thx in advance
There is a simple way in OpenCV using findContours() and drawContours(). If you use the hierarchical version of findContours(), you can then look through the hierarchy and draw (fill) the child contour of the white quad only. This has the additional advantage that you can do some sanity checks (e.g. checking the size of the contour to see if it is approximately the size you expect) if necessary. I don't know anything about java or javacv, but maybe you can check out the c++ example for findcontours included with opencv for inspiration?
You can detect image objects on images by using openCV library(through java adapter); for it you will need to train network for circles.
Regarding exactly your case(probably this solution will not be generic) you can split your image on segments, and using as condition - color changing, see pseudo code below:
//build color switching list
List<Point> colorSwitches = ...
for(each horizontal line from image){
for(each pixel from line){
if(color of previous pixel != color of current pixel){
colorSwitches.add(currentPoint)
}
}
}
// so, you have detected margins of your image objects; now we need to merge neighbor pixels into image objects, where image object is collection of margin points(you should create this class)
List<ImageObject> imageObjects = ...
for(each color switch){
if(current pixel is connected with pixels from existing image objects){
// returns image object neared with current point
getRelatedImageObject(imageObjects).add(currentPoint);
}else{
imageObjects.add(new ImageObject(currentPixel));
}
}
// now we have list of objects from image, and we need to match objects
Ok, its general lines how to do what you need, if you will need more exactly I will try to explain more detailed. Also you can contact me directly
Hope it will help you.