I am currently working the image to transform the sector part in the image to a rectangle shape.
I tried some ways but don't work well.
Any ideas on how to transform?
Create a destination image the height of which is the difference in the radii, and the width is the length of the perimeter at middle radius (this ensures square pixels along the middle arc).
Scan this image and for every pixel convert the coordinates (angle, radius) from Cartesian to polar, with a shift to the ROI center. This will give you the corresponding pixel in the source image, which you copy to the destination. Make sure to scale the angle and radius to match the destination image limits to the ROI edges.
As the source coordinates won't be integer, truncating and merely copying the source pixel achieves so-called nearest-neighbor resampling, which features visible artifacts. You can smoothen them by considering four neighboring pixels and interpolating bilinearly between them by means of the fractional parts of the coordinates.
You can even go for bicubic interpolation, using sixteen neighbors. But from my experience, the gain in quality is not so visible.
Related
Currently I am working with a sudoku grid and I have the binary image. I am using Regionprops to get the area of the connected components and then turn the rest of the image black. After this I call the OCR method to try and read the sudoku numbers. The problem is that this only works if the sudoku grid in the image is straight and upright. If it is rotated even a little bit I am not able to pull the numbers. This is the code I have so far:
% get grid connected parts
conn_part = bwconncomp(im_binary);
% blacken area outside
stats = regionprops(conn_part,'Area');
im_out = im_binary; % Make mask
im_out(vertcat(conn_part.PixelIdxList{[stats.Area] < 825 | [stats.Area] > 2500})) = 0;
imagesc(im_out);
title("Numbers pulled");
sudokuNum = ocr(im_out,'TextLayout','Block','CharacterSet','0123456789');
sudokuNum.Text;
Where im_binary is the binary image
im_out is the output image
stats is the object returned from regionprops containing the area of the connected components
I know I can rotate the image before getting the OCR results by doing:
im_out = imrotate(im_out, angle)
However I don't know what angle the grid is at since this is part of a function that loops through for multiple images. I looked into the regionprops method because there is an attribute 'Orientation' which I can pull from there but I don't understand how I would actually use it. It also states that regionprops will return a value between -90 and 90, but my image could be rotated by more than 90 degrees.
Don't rotate the connected component or the binary image. First use the binary image to determine the rotation, then rotate the original grey-scale or color input image, and then binarize the rotated image. You'll be able to transform with interpolation, which will improve your results greatly. It does require to do the binarization step twice, but I don't think this step usually is too expensive.
The regionprops orientation feature is computed by "fitting" an ellipse to the shape. This is meaningful only for elongated objects. For a square sudoku grid this will not yield any valuable information.
Instead, look at the angle at which the smallest Feret diameter was obtained. The Feret diameters are the lengths of the projections at arbitrary angles. At one angle, this projection is smallest. By necessity it will be at an angle corresponding to one of the principal axes of the square. Here is more information about how to compute Feret diameters in MATLAB.
A different alternative is e.g. to use the Hough transform to detect the lines of the grid.
Do note that the geometry of the puzzle will never tell you about which side is up. The angle you get here should be taken modulo π/2 (i.e. constrain to the range -π/4 to π/4).
To know what direction is up you might do by trying to read the text, if it fails, rotate by 90 degrees and try again.
I am currently taking my first steps in the field of computer vision and image processing.
One of the tasks I'm working on is finding the center coordinates of (overlapping and occluded) circles.
Here is a sample image:
Here is another sample image showing two overlapping circles:
Further information about the problem:
Always a monochrome, grayscale image
Rather low resolution images
Radii of the circles are unknown
Number of circles in a given image is unknown
Center of circle is to be determined, preferably with sub-pixel accuracy
Radii do not have to be determined
Relative low overhead of the algorithm is of importance; the processing is supposed to be carried out with real-time camera images
For the first sample image, it is relatively easy to calculate the center of the circle by finding the center of mass. Unfortunately, this is not going to work for the second image.
Things I tried are mainly based on the Circle Hough Transform and the Distance Transform.
The Circle Hough Transform seemed relatively computationally expensive due to the fact that I have no information about the radii and the range of possible radii is large. Furthermore, it seems hard to identify the (appropriate) pixels along the edge because of the low resolution of the image.
As for the Distance Transform, I have trouble identifying the centers of the circles and the fact that the image needs to be binarized implies a certain loss of information.
Now I am looking for viable alternatives to the aforementioned algorithms.
Some more sample images (images like the two samples above are extracted from images like the following):
Just thinking aloud to try and get the ball rolling for you... I would be thinking of a Blob, or Connected Component analysis to separate out your blobs.
Then I would start looking at each blob individually. First thing is to see how square the bounding box is for each blob. If it is pretty square AND the centroid of the blob is central within the square, then you have a single circle. If it is not square, or the centroid is not central, you have more than one circle.
Now I am going to start looking at where the white areas touch the edges of the bounding box for some clues as to where the centres are...
Hy!
I have a point cloud representing walls and the floor of an indoor scene.
I projected the points of the floor on a plane. That means it's a "2d point cloud" now.
All z-coordinates are zero.
I have to deal with missing parts. My idea now is to fill the holes.
Is there a way to transform the points into the image space to create a
binary mask? I would like to use techniques like: "imfill" in Matlab...
Thanks
Edit:
To make it more clear, I will explain an simple example. I have points in 2D. After I made a triangulation, I can access each triangle. For each triangle I create a binary mask with poly2mask(), and each mask I write to an final image.
Here is an example:
Now I can use morphological operations on the image.
E.G: Here is an more complex example, where the triangulation gives me bad results:
To fill the hole on the right side, I could use morphological operation.
My problem: The points can be negative, and the distance between the triangles can be very small (E.g.: x coordinates of triangle: (1.12 1.14 1.12), will give me the point 1 in the image space
Simple rounded corner rectangle code in Matlab can be written as follows.
rectangle('Position',[0,-1.37/2,3.75,1.37],...
'Curvature',[1],...
'LineWidth',1,'LineStyle','-')
daspect([1,1,1])
How to get the x and y coordinates arrays of this figure?
To get the axes units boundaries, do:
axisUnits = axis(axesHandle) % axesHandle could be gca
axisUnits will be an four elements array, with the following syntax: [xlowlim xhighlim ylowlim yhighlim], it will also contain the zlow and zhigh for 3-D plots.
But I think that is not what you need to know. Checking the matlab documentation for the rectangle properties, we find:
Position four-element vector [x,y,width,height]
Location and size of rectangle. Specifies the location and size of the
rectangle in the data units of the axes. The point defined by x, y
specifies one corner of the rectangle, and width and height define the
size in units along the x- and y-axes respectively.
It is also documented on the rectangle documentation:
rectangle('Position',[x,y,w,h]) draws the rectangle from the point x,y
and having a width of w and a height of h. Specify values in axes data
units.
See if this illustrate what you want. You have an x axis that goes from −100 to 100 and y axis that goes from 5 to 15. Suppose you want to put a rectangle from −30 to −20 in x and 8 to 10 in y.
rectangle('Position',[-30,8,10,2]);
As explained by the comments there appears to be no direct way to query the figure created by rectangle and extract x/y coordinates. On the other hand, I can think of two simple strategies to arrive at coordinates that will closely reproduce the curve generated with rectangle:
(1) Save the figure as an image (say .png) and process the image to extract points corresponding to the curve. Some degree of massaging is necessary but this is relatively straightforward if blunt and I expect the code to be somewhat slow at execution compared to getting data from an axes object.
(2) Write your own code to draw a rectangle with curved edges. While recreating precisely what matlab draws may not be so simple, you may be satisfied with your own version.
Whether you choose one of these approaches boils down to (a) what speed of execution you consider acceptable (b) how closely you need to replicate what rectangle draws on screen (c) whether you have image processing routines, say for reading an image file.
Edit
If you have the image processing toolbox you can arrive at a set of points representing the rectangle as follows:
h=rectangle('Position',[0,-1.37/2,3.75,1.37],...
'Curvature',[1],...
'LineWidth',1,'LineStyle','-')
daspect([1,1,1])
axis off
saveas(gca,'test.png');
im = imread('test.png');
im = rgb2gray(im);
figure, imshow(im)
Note that you will still need to apply a threshold to pick the relevant points from the image and then transform the coordinate system and rearrange the points in order to display properly as a connected set. You'll probably also want to tinker with resolution of the initial image file or apply image processing functions to get a smooth curve.
I need to crop a circle in MATLAB.
I need to perform iris segmentation, and I´ve identified the center point and the radius of the iris, and I need to cut it off from the image.
I have a vector ci that ci(1) is X-coordinate ci(2) is Y-coordinate and ci(3) is the radius of the circle.
One way to do this is to create a binary mask with ones inside the circle and zeros outside. You can then use this array to either mask everything outside the circle with NaNs, or to read the pixel values of the image inside the mask.
To create a circle mask, an easy way is to create coordinate arrays centered on the iris, and threshold the distance, like this:
[xx,yy] = ndgrid((1:imageSize(1))-ci(1),(1:imageSize(2))-ci(2));
mask = (xx.^2 + yy.^2)<ci(3)^2;