This is a continuation of this question : Finding Squares in Image
I followed the steps in my answer there : https://dsp.stackexchange.com/a/7526/818, And I got the answer as given below :
But at the end of that answer, I have explained a problem, and that is my question.
Explanation :
I already have the centroids of detected squares from step 1 in the previous link (and those detected squares are marked in mask_image below):
I created a grid image as below ( I know their centroid values also) :
I also found which point in grid image to be mapped to corresponding point in mask_image.
With that information, I applied scipy.interpolate.griddata(), and then OpenCV's cv2.remap() function.
And its result is given below :
As you can see, all the squares except the two at center are clipped. It is like, output contain only the region inside a boundary drawn connecting all the centroids of the mask_image.
Scene becomes more worse below :
The case is more worse when the last square (yellow) or any other square in four corners is not detected in first step. Consider last one is not detected. Then below is the result I get and you can see a slant cut at the bottom (marked with yellow color):
Question :
Why remap function not working beyond the points I have given ? And what should I do to remap it without clipping ?
I thought it would work for the full image even if I give some points, which are not at edges.
Expected Output :
Below is the output I expected at the end of my operation. (Region inside red boundary is what I got actually got now )
Looking for some good suggestions ...
UPDATE :
I also add the code here. Only remapping part code is added. Full code is too big to be added here :
# ideal - the grid image - http://i.stack.imgur.com/3QudG.png
# centroids - list of centroids of the squares in mask_image - http://i.stack.imgur.com/jh6bQ.png
# match_pts - list of centroids of the squares in grid image corresponding to squares in mask_image
# warped - the final image obtained after remap - http://i.stack.imgur.com/O26ZA.png
grid_x,grid_y = np.meshgrid(np.arange(ideal.shape[1]),np.arange(ideal.shape[0]))
dst = np.array(centroids)
src = np.array(match_pts)
grid_z = griddata(dst,src,(grid_x,grid_y),method='cubic')
map_x_32 = grid_z[:,:,0].astype('float32')
map_y_32 = grid_z[:,:,1].astype('float32')
warped = cv2.remap(ideal, map_x_32, map_y_32, cv2.INTER_CUBIC)
Also added the datas like centroids, match_pts etc, so that if someone want to try it out, they can directly use the data instead of finding it from image : gist.github.com/4540887
Step 1: Whatever final binary image you are getting from analyzing in B,G,R,H,S,V plane, in that image do a blob counting algorithm.
Step 2: Find the largest blob on basis of area or contour length. Since your blobs will be mostly parallelogram types so area or contour, any one will do.
Step 3: With the largest blob (since largest blob is the best blob resembling your real world squares) try to find the orientation of the blob...this you can get by a fitting a best fit rectangle OR you can get the corner points...get the slope of the lines joining them (in both horizon and vertical direction).
Step 4: Once you get the two slopes draw two lines running through the axis of the blob. for axis you can average the corner points or you can use the centroid (center of mass)...I would go with average of corner points...
Step 5: Since in each horizontal and vertical direction, spacing is equal (ideally horizontal and vertical spacing are also equal as it comes from your ideal square picture but we will not assume it..) just need to locate the possible centroids of the other parallelograms
BOTTOM LINE: If any one square gets detected perfectly you can make the whole grid. Just keep marking centers at an interval of 2H (H = horizontal width of biggest blob) along the horizontal axis of the biggest blob and at an interval of 2V (V = vertical height of biggest blob) vertically along the vertical axis of the blob.
Some pics to support
Related
I have a problem where we have a grid of points and I'd like to fit a "deformed grid which would best fit the set of points.
The MatLab data can be found at:
https://drive.google.com/file/d/14fKKEC5BKGDOjzWupWFSmythqUrRXae4/view?usp=sharing
You will see that cenX and cenY are the x and y coordinates of these centroids.
Like on this image. To note is that there are points missing, and there are a few extra points. Moreover, You can see some lines are not one single line from left to right, however, we could safely assume that the fitting a line somewhat horizontally (+-5degrees) would properly link the points into a somewhat deformed grid.
The vertical lines are trivial because that is how we generated these dots. We can find the number of lines required through a mode of the count of points on each of the columns of the grid.
I'd like to be able to ensure that a point is only part of one line, as this is a grid.
I was working on my image processing problem with detecting coins.
I have some images like this one here:
and wanted to separate the falsely connected coins.
We already tried the watershed method as stated on the MATLAB-Homepage:
the-watershed-transform-strategies-for-image-segmentation.html
especially since the first example is exactly our problem.
But instead we get a somehow very messed up separation as you can see here:
We already extracted the area of the coin using the regionprops Extrema parameter and casting the watershed only on the needed area.
I'd appreciate any help with the problem or even another method of getting it separated.
If you have the Image Processing Toolbox, I can also suggest the Circular Hough Transform through imfindcircles. However, this requires at least version R2012a, so if you don't have it, this won't work.
For the sake of completeness, I'll assume you have it. This is a good method if you want to leave the image untouched. If you don't know what the Hough Transform is, it is a method for finding straight lines in an image. The circular Hough Transform is a special case that aims to find circles in the image.
The added advantage of the circular Hough Transform is that it is able to detect partial circles in an image. This means that those regions in your image that are connected, we can detect them as separate circles. How you'd call imfindcircles is in the following fashion:
[centers,radii] = imfindcircles(A, radiusRange);
A would be your binary image of objects, and radiusRange is a two-element array that specifies the minimum and maximum radii of the circles you want to detect in your image. The outputs are:
centers: A N x 2 array that tells you the (x,y) co-ordinates of each centre of a circle that is detected in the image - x being the column and y being the row.
radii: For each corresponding centre detected, this also gives the radius of each circle detected. This is a N x 1 array.
There are additional parameters to imfindcircles that you may find useful, such as the Sensitivity. A higher sensitivity means that it is able to detect circular shapes that are more non-uniform, such as what you are showing in your image. They aren't perfect circles, but they are round shapes. The default sensitivity is 0.85. I set it to 0.9 to get good results. Also, playing around with your image, I found that the radii ranged from 50 pixels to 150 pixels. Therefore, I did this:
im = im2bw(imread('http://dennlinger.bplaced.net/t06-4.jpg'));
[centers,radii] = imfindcircles(im, [50 150], 'Sensitivity', 0.9);
The first line of code reads in your image directly from StackOverflow. I also convert this to logical or true black and white as the image you uploaded is of type uint8. This image is stored in im. Next, we call imfindcircles in the method that we described.
Now, if we want to visualize the detected circles, simply use imshow to show your image, then use the viscircles to draw the circles in the image.
imshow(im);
viscircles(centers, radii, 'DrawBackgroundCircle', false);
viscircles by default draws the circles with a white background over the contour. I want to disable this because your image has white circles and I don't want to show false contouring. This is what I get with the above code:
Therefore, what you can take away from this is the centers and radii variables. centers will give you the centre of each detected circle while radii will tell you what the radii is for each circle.
Now, if you want to simulate what regionprops is doing, we can iterate through all of the detected circles and physically draw them onto a 2D map where each circle would be labeled by an ID number. As such, we can do something like this:
[X,Y] = meshgrid(1:size(im,2), 1:size(im,1));
IDs = zeros(size(im));
for idx = 1 : numel(radii)
r = radii(idx);
cen = centers(idx,:);
loc = (X - cen(1)).^2 + (Y - cen(2)).^2 <= r^2;
IDs(loc) = idx;
end
We first define a rectangular grid of points using meshgrid and initialize an IDs array of all zeroes that is the same size as the image. Next, for each pair of radii and centres for each circle, we define a circle that is centered at this point that extends out for the given radius. We then use these as locations into the IDs array and set it to a unique ID for that particular circle. The result of IDs will be that which resembles the output of bwlabel. As such, if you want to extract the locations of where the idx circle is, you would do:
cir = IDs == idx;
For demonstration purposes, this is what the IDs array looks like once we scale the IDs such that it fits within a [0-255] range for visibility:
imshow(IDs, []);
Therefore, each shaded circle of a different shade of gray denotes a unique circle that was detected with imfindcircles.
However, the shades of gray are probably a bit ambiguous for certain coins as this blends into the background. Another way that we could visualize this is to apply a different colour map to the IDs array. We can try using the cool colour map, with the total number of colours to be the number of unique circles + 1 for the background. Therefore, we can do something like this:
cmap = cool(numel(radii) + 1);
RGB = ind2rgb(IDs, cmap);
imshow(RGB);
The above code will create a colour map such that each circle gets mapped to a unique colour in the cool colour map. The next line applies a mapping where each ID gets associated with a colour with ind2rgb and we finally show the image.
This is what we get:
Edit: the following solution is more adequate to scenarios where one does not require fitting the exact circumferences, although simple heuristics could be used to approximate the radii of the coins in the original image based on the centers found in the eroded one.
Assuming you have access to the Image Processing toolbox, try imerode on your original black and white image. It will apply an erosion morphological operator to your image. In fact, the Matlab webpage with the documentation of that function has an example strikingly similar to your problem/image and they use a disk structure.
Run the following code (based on the example linked above) assuming the image you submitted is called ima.jpg and is local to the code:
ima=imread('ima.jpg');
se = strel('disk',50);
eroded = imerode(ima,se);
imshow(eroded)
and you will see the image that follows as output. After you do this, you can use bwlabel to label the connected components and compute whatever properties you may want, for example, count the number of coins or detect their centers.
How do I separate the two connected circles in the image below, using MATLAB? I have tried using imerode, but this does not give good results. Eroding does not work, because in order to erode enough to separate the circles, the lines disappear or become mangled. In other starting pictures, a circle and a line overlap, so isolating the overlapping objects won't work either.
The image shows objects identified by bwboundaries, each object painted a different color. As you can see, the two light blue circles are joined, and I want to disjoin them, producing two separate circles. Thanks
I would recommend you use the Circular Hough Transform through imfindcircles. However, you need version 8 of the Image Processing Toolbox, which was available from version R2012a and onwards. If you don't have this, then unfortunately this won't work :(... but let's go with the assumption that you do have it. However, if you are using something older than R2012a, Dev-iL in his/her comment above linked to some code on MATLAB's File Exchange on an implementation of this, most likely created before the Circular Hough Transform was available: http://www.mathworks.com/matlabcentral/fileexchange/9168-detect-circles-with-various-radii-in-grayscale-image-via-hough-transform/
This is a special case of the Hough Transform where you are trying to find circles in your image rather than lines. The beauty with this is that you are able to find circles even when the circle is partially completed or overlapping.
I'm going to take the image that you provided above and do some post-processing on it. I'm going to convert the image to binary, and remove the border, which is white and contains the title. I'm also going to fill in any holes that result so that all of the objects are filled in with solid white. There is also some residual quantization noise after I do this step, so I'm going to a small opening with a 3 x 3 square element. After, I'm going to close the shapes with a 3 x 3 square element, as I see that there are noticeable gaps in the shapes. Therefore:
Therefore, directly reading in your image from where you've posted it:
im = imread('http://s29.postimg.org/spkab8oef/image.jpg'); %// Read in the image
im_gray = im2double(rgb2gray(im)); %// Convert to grayscale, then [0,1]
out = imclearborder(im_gray > 0.6); %// Threshold using 0.6, then clear the border
out = imfill(out, 'holes'); %// Fill in the holes
out = imopen(out, strel('square', 3));
out = imclose(out, strel('square', 3));
This is the image I get:
Now, apply the Circular Hough Transform. The general syntax for this is:
[centres, radii, metric] = imfindcircles(img, [start_radius, end_radius]);
img would be the binary image that contains your shapes, start_radius and end_radius would be the smallest and largest radius of the circles you want to find. The Circular Hough Transform is performed such that it will find any circles that are within this range (in pixels). The outputs are:
centres: Which returns the (x,y) positions of the centres of each circle detected
radii: The radius of each circle
metric: A measure of purity of the circle. Higher values mean that the shape is more probable to be a circle and vice-versa.
I searched for circles having a radius between 30 and 60 pixels. Therefore:
[centres, radii, metric] = imfindcircles(out, [30, 60]);
We can then demonstrate the detected circles, as well as the radii by a combination of plot and viscircles. Therefore:
imshow(out);
hold on;
plot(centres(:,1), centres(:,2), 'r*'); %// Plot centres
viscircles(centres, radii, 'EdgeColor', 'b'); %// Plot circles - Make edge blue
Here's the result:
As you can see, even with the overlapping circles towards the top, the Circular Hough Transform was able to detect two distinct circles in that shape.
Edit - November 16th, 2014
You wish to ensure that the objects are separated before you do bwboundaries. This is a bit tricky to do. The only way I can see you do this is if you don't even use bwboundaries at all and do this yourself. I'm assuming you'll want to analyze each shape's properties by themselves after all of this, so what I suggest you do is iterate through every circle you have, then place each circle on a new blank image, do a regionprops call on that shape, then append it to a separate array. You can also keep track of all of the circles by having a separate array that adds the circles one at a time to this array.
Once you've finished with all of the circles, you'll have a structure array that contains all of the measured properties for all of the measured circles you have found. You would use the array that contains only the circles from above, then use these and remove them from the original image so you get just the lines. You'd then call one more regionprops on this image to get the information for the lines and append this to your final structure array.
Here's the first part of the procedure I outlined above:
num_circles = numel(radii); %// Get number of circles
struct_reg = []; %// Save the shape analysis per circle / line here
%// For creating our circle in the temporary image
[X,Y] = meshgrid(1:size(out,2), 1:size(out,1));
%// Storing all of our circles in this image
circles_img = false(size(out));
for idx = 1 : num_circles %// For each circle we have...
%// Place our circle inside a temporary image
r = radii(idx);
cx = centres(idx,1); cy = centres(idx,2);
tmp = (X - cx).^2 + (Y - cy).^2 <= r^2;
% // Save in master circle image
circles_img(tmp) = true;
%// Do regionprops on this image and save
struct_reg = [struct_reg; regionprops(tmp)];
end
The above code may be a bit hard to swallow, but let's go through it slowly. I first figure out how many circles we have, which is simply looking at how many radii we have detected. I keep a separate array called struct_reg that will append a regionprops struct for each circle and line we have in our image. I use meshgrid to determine the (X,Y) co-ordinates with respect to the image containing our shapes so that I can draw one circle onto a blank image at each iteration. To do this, you simply need to find the Euclidean distance with respect to the centre of each circle, and set the pixels to true only if that location has its distance less than r. After doing this operation, you will have created only one circle and filtered all of them out. You would then use regionprops on this circle, add it to our circles_img array, which will only contain the circles, then continue with the rest of the circles.
At this point, we will have saved all of our circles. This is what circles_img looks like so far:
You'll notice that the circles drawn are clean, but the actual circles in the original image are a bit jagged. If we tried to remove the circles with this clean image, you will get some residual pixels along the border and you won't completely remove the circles themselves. To illustrate what I mean, this is what your image looks like if I tried to remove the circles with circles_img by itself:
... not good, right?
If you want to completely remove the circles, then do a morphological reconstruction through imreconstruct where you can use this image as the seed image, and specify the original image to be what we're working on. The job of morphological reconstruction is essentially a flood fill. You specify seed pixels, and an image you want to work on, and the job of imreconstruct is from these seeds, flood fill with white until we reach the boundaries of the objects that the seed pixels resided in. Therefore:
out_circles = imreconstruct(circles_img, out);
Therefore, we get this for our final reconstructed circles image:
Great! Now, use this and remove the circles from the original image. Once you do this, run regionprops again on this final image and append to your struct_reg variable. Obviously, save a copy of the original image before doing this:
out_copy = out;
out_copy(out_circles) = false;
struct_reg = [struct_reg; regionprops(out_copy)];
Just for sake of argument, this is what the image looks like with the circles removed:
Now, we have analyzed all of our shapes. Bear in mind I did the full regionprops call because I don't know exactly what you want in your analysis... so I just decided to give you everything.
Hope this helps!
erosion is the way to go. You should probably use a larger structuring element.
How about
1 erode
2 detect your objects
3 dilate each object for itself using the same structuring element
Consider that I have a colored image like this in which the outline is not complete (There are gaps between lines). I want to be able to fill the area between the lines with one color or another. This actually is a binary image which I got after applying canny edge detector on a corresponding gray scale image.
I tried first dilating the image and then eroding it, but the result is not good enough. I want to be able to preserve the thickness of the root
Any help would be greatly appreciated
Original Image
Image after edge detection and some manual removal of pixels
Using the information in the edge image, I thought I would try to extract pixels from the original image of a certain color. For every white pixel in the edited image, I used a search space in the original image along the same horizontal line. I used different thresholds for R, G and B and I ended up with this
I'm not sure what your original image looks like. It would be helpful to see.
You have gaps between the lines because a line in your original image has two edges, one on each side. The canny algorithm is detecting them both. The Canny edge detection algorithm has at its heart the application of two Sobel kernels to calculate the gradient, one for detecting horizontal edges and one for detection vertical edges.
-1 0 +1
-2 0 +2
-1 0 +1
and
+1 +2 +1
0 0 0
-1 -2 -1
These kernels will present peaks for both sides of the line. One peak positive and one negative. You can exclude one side of the line by excluding the corresponding peak. After taking the gradient of each direction truncate any values below zero (set the values to zero) to remove the second peak. Then continue with the Canny edge detection as usual. This will result in the detection of only a single edge for each line instead of the two that you are seeing now.
I'll add a third approach now that I have seen the image. It looks like most of the information is in the green channel.
Green channel image
This image gives you a decent result if you simply apply a threshold.
Thresholded image with a somewhat arbitrary threshold
You can then either clean this image up by itself or use your edge image. To clean it up with the edge image you produced remove any white pixels that are more than a certain distance from one of your detected edges (create a Euclidean distance map from your edge image and use that to set any white pixels greater than a certain distance from an edge to black).
If you are still collecting images you may want to try to position the camera in a way to avoid the bottom of the jar (or whatever this is).
You could attempt to use a line scanning methodology. Start at the side and scan horizontally. When you hit an edge you assume you are in a root and you start setting the voxels to white. When you hit another edge you assume you are leaving a root and you start. There will be some fringe cases and you may want to add additional checks, such as limiting the allowed thickness of a root.
You could also do a flood fill style algorithm where you take a seed point in a root and travel up the root filling it in.
Not sure how well these would work as it depends on the image and I did not test it.
i need to find the distance between the two points.I can find the distance between them manually by the pixel to cm converter in the image processing tool box. But i want a code which detects the point positions in the image and calculate the distance.
More accurately speaking the image contains only three points one mid and the other two approximately distanced equally from it...
There might be a better way then this, but I hacked something similar together last night.
Use bwboundaries to find the objects in the image (the contiguous regions in a black/white image).
The second returned matrix, L, is the same image but with the regions numbered. So for the first point, you want to isolate all the pixels related to it,
L2 = (L==1)
Now find the center of that region (for object 1).
x1 = (1:size(L2,2))*sum(L2,1)'/size(L2,2);
y1 = (1:size(L2,1))*sum(L2,2)/size(L2,1);
Repeat that for all the regions in your image. You should have the center of mass of each point. I think that should do it for you, but I haven't tested it.