I am working on images to detect and fit the largest possible circle in any of the free areas of an image containing distributed particles:
(able to detect the location of particle).
One direction is to define a circle touching any 3-point combination, checking if the circle is empty, then finding the largest circle among all empty circles. However, it leads to a huge number of combination i.e. C(n,3), where n is the total number of particles in the image.
I would appreciate if anyone can provide me any hint or alternate method that I can explore.
Lets do some maths my friend, as maths will always get to the end!
Wikipedia:
In mathematics, a Voronoi diagram is a partitioning of a plane into
regions based on distance to points in a specific subset of the plane.
For example:
rng(1)
x=rand(1,100)*5;
y=rand(1,100)*5;
voronoi(x,y);
The nice thing about this diagram is that if you notice, all the edges/vertices of those blue areas are all to equal distance to the points around them. Thus, if we know the location of the vertices, and compute the distances to the closest points, then we can choose the vertex with highest distance as our center of the circle.
Interestingly, the edges of a Voronoi regions are also defined as the circumcenters of the triangles generated by a Delaunay triangulation.
So if we compute the Delaunay triangulation of the area, and their circumcenters
dt=delaunayTriangulation([x;y].');
cc=circumcenter(dt); %voronoi edges
And compute the distances between the circumcenters and any of the points that define each triangle:
for ii=1:size(cc,1)
if cc(ii,1)>0 && cc(ii,1)<5 && cc(ii,2)>0 && cc(ii,2)<5
point=dt.Points(dt.ConnectivityList(ii,1),:); %the first one, or any other (they are the same distance)
distance(ii)=sqrt((cc(ii,1)-point(1)).^2+(cc(ii,2)-point(2)).^2);
end
end
Then we have the center (cc) and radius (distance) of all possible circles that have no point inside them. We just need the biggest one!
[r,ind]=max(distance); %Tada!
Now lets plot
hold on
ang=0:0.01:2*pi;
xp=r*cos(ang);
yp=r*sin(ang);
point=cc(ind,:);
voronoi(x,y)
triplot(dt,'color','r','linestyle',':')
plot(point(1)+xp,point(2)+yp,'k');
plot(point(1),point(2),'g.','markersize',20);
Notice how the center of the circle is on one vertex of the Voronoi diagram.
NOTE: this will find the center inside [0-5],[0-5]. you can easily modify it to change this constrain. You can also try to find the circle that fits on its entirety inside the interested area (as opposed to just the center). This would require a small addition in the end where the maximum is obtained.
I'd like to propose another solution based on a grid search with refinement. It's not as advanced as Ander's or as short as rahnema1's, but it should be very easy to follow and understand. Also, it runs quite fast.
The algorithm contains several stages:
We generate an evenly-spaced grid.
We find the minimal distances of points in the grid to all provided points.
We discard all points whose distances are below a certain percentile (e.g. 95th).
We choose the region which contains the largest distance (this should contain the correct center if my initial grid is fine enough).
We create a new meshgrid around the chosen region and find distances again (this part is clearly sub-optimal, because the distances are computed to all points, including far and irrelevant ones).
We iterate the refinement within the region, while keeping an eye on the variance of the top 5% of values -> if it drops below some preset threshold we break.
Several notes:
I have made the assumption that circles cannot go beyond the scattered points' extent (i.e. the bounding square of the scatter acts as an "invisible wall").
The appropriate percentile depends on how fine the initial grid is. This will also affect the amount of while iterations, and the optimal initial value for cnt.
function [xBest,yBest,R] = q42806059
rng(1)
x=rand(1,100)*5;
y=rand(1,100)*5;
%% Find the approximate region(s) where there exists a point farthest from all the rest:
xExtent = linspace(min(x),max(x),numel(x));
yExtent = linspace(min(y),max(y),numel(y)).';
% Create a grid:
[XX,YY] = meshgrid(xExtent,yExtent);
% Compute pairwise distance from grid points to free points:
D = reshape(min(pdist2([XX(:),YY(:)],[x(:),y(:)]),[],2),size(XX));
% Intermediate plot:
% figure(); plot(x,y,'.k'); hold on; contour(XX,YY,D); axis square; grid on;
% Remove irrelevant candidates:
D(D<prctile(D(:),95)) = NaN;
D(D > xExtent | D > yExtent | D > yExtent(end)-yExtent | D > xExtent(end)-xExtent) = NaN;
%% Keep only the region with the largest distance
L = bwlabel(~isnan(D));
[~,I] = max(table2array(regionprops('table',L,D,'MaxIntensity')));
D(L~=I) = NaN;
% surf(XX,YY,D,'EdgeColor','interp','FaceColor','interp');
%% Iterate until sufficient precision:
xExtent = xExtent(~isnan(min(D,[],1,'omitnan')));
yExtent = yExtent(~isnan(min(D,[],2,'omitnan')));
cnt = 1; % increase or decrease according to the nature of the problem
while true
% Same ideas as above, so no explanations:
xExtent = linspace(xExtent(1),xExtent(end),20);
yExtent = linspace(yExtent(1),yExtent(end),20).';
[XX,YY] = meshgrid(xExtent,yExtent);
D = reshape(min(pdist2([XX(:),YY(:)],[x(:),y(:)]),[],2),size(XX));
D(D<prctile(D(:),95)) = NaN;
I = find(D == max(D(:)));
xBest = XX(I);
yBest = YY(I);
if nanvar(D(:)) < 1E-10 || cnt == 10
R = D(I);
break
end
xExtent = (1+[-1 +1]*10^-cnt)*xBest;
yExtent = (1+[-1 +1]*10^-cnt)*yBest;
cnt = cnt+1;
end
% Finally:
% rectangle('Position',[xBest-R,yBest-R,2*R,2*R],'Curvature',[1 1],'EdgeColor','r');
The result I'm getting for Ander's example data is [x,y,r] = [0.7832, 2.0694, 0.7815] (which is the same). The execution time is about half of Ander's solution.
Here are the intermediate plots:
Contour of the largest (clear) distance from a point to the set of all provided points:
After considering distance from the boundary, keeping only the top 5% of distant points, and considering only the region which contains the largest distance (the piece of surface represents the kept values):
And finally:
You can use bwdist from Image Processing Toolbox to compute the distance transform of the image. This can be regarded as a method to create voronoi diagram that well explained in #AnderBiguri's answer.
img = imread('AbmxL.jpg');
%convert the image to a binary image
points = img(:,:,3)<200;
%compute the distance transform of the binary image
dist = bwdist(points);
%find the circle that has maximum radius
radius = max(dist(:));
%find position of the circle
[x y] = find(dist == radius);
imshow(dist,[]);
hold on
plot(y,x,'ro');
The fact that this problem can be solved using a "direct search" (as can be seen in another answer) means one can look at this as a global optimization problem. There exist various ways to solve such problems, each appropriate for certain scenarios. Out of my personal curiosity I have decided to solve this using a genetic algorithm.
Generally speaking, such an algorithm requires us to think of the solution as a set of "genes" subject to "evolution" under a certain "fitness function". As it happens, it's quite easy to identify the genes and the fitness function in this problem:
Genes: x , y, r.
Fitness function: technically, maximum area of circle, but this is equivalent to the maximum r (or minimum -r, since the algorithm requires a function to minimize).
Special constraint - if r is larger than the euclidean distance to the closest of the provided points (that is, the circle contains a point), the organism "dies".
Below is a basic implementation of such an algorithm ("basic" because it's completely unoptimized, and there is lot of room for optimizationno pun intended in this problem).
function [x,y,r] = q42806059b(cloudOfPoints)
% Problem setup
if nargin == 0
rng(1)
cloudOfPoints = rand(100,2)*5; % equivalent to Ander's initialization.
end
%{
figure(); plot(cloudOfPoints(:,1),cloudOfPoints(:,2),'.w'); hold on; axis square;
set(gca,'Color','k'); plot(0.7832,2.0694,'ro'); plot(0.7832,2.0694,'r*');
%}
nVariables = 3;
options = optimoptions(#ga,'UseVectorized',true,'CreationFcn',#gacreationuniform,...
'PopulationSize',1000);
S = max(cloudOfPoints,[],1); L = min(cloudOfPoints,[],1); % Find geometric bounds:
% In R2017a: use [S,L] = bounds(cloudOfPoints,1);
% Here we also define distance-from-boundary constraints.
g = ga(#(g)vectorized_fitness(g,cloudOfPoints,[L;S]), nVariables,...
[],[], [],[], [L 0],[S min(S-L)], [], options);
x = g(1); y = g(2); r = g(3);
%{
plot(x,y,'ro'); plot(x,y,'r*');
rectangle('Position',[x-r,y-r,2*r,2*r],'Curvature',[1 1],'EdgeColor','r');
%}
function f = vectorized_fitness(genes,pts,extent)
% genes = [x,y,r]
% extent = [Xmin Ymin; Xmax Ymax]
% f, the fitness, is the largest radius.
f = min(pdist2(genes(:,1:2), pts, 'euclidean'), [], 2);
% Instant death if circle contains a point:
f( f < genes(:,3) ) = Inf;
% Instant death if circle is too close to boundary:
f( any( genes(:,3) > genes(:,1:2) - extent(1,:) | ...
genes(:,3) > extent(2,:) - genes(:,1:2), 2) ) = Inf;
% Note: this condition may possibly be specified using the A,b inputs of ga().
f(isfinite(f)) = -genes(isfinite(f),3);
%DEBUG:
%{
scatter(genes(:,1),genes(:,2),10 ,[0, .447, .741] ,'o'); % All
z = ~isfinite(f); scatter(genes(z,1),genes(z,2),30,'r','x'); % Killed
z = isfinite(f); scatter(genes(z,1),genes(z,2),30,'g','h'); % Surviving
[~,I] = sort(f); scatter(genes(I(1:5),1),genes(I(1:5),2),30,'y','p'); % Elite
%}
And here's a "time-lapse" plot of 47 generations of a typical run:
(Where blue points are the current generation, red crosses are "insta-killed" organisms, green hexagrams are the "non-insta-killed" organisms, and the red circle marks the destination).
I'm not used to image processing, so it's just an Idea:
Implement something like a gaussian filter (blur) which transforms each particle (pixels) to a round gradiant with r=image_size (all of them overlapping). This way, you should get a picture where the most white pixels should be the best results. Unfortunately, the demonstration in gimp failed because the extreme blurring made the dots disappearing.
Alternatively, you could incrementelly extend all existing pixels by marking all neighbour pixels in an area (example: r=4), the pixels left would be the same result (those with the biggest distance to any pixel)
I have a binary image of a human. In MATLAB, boundary points and the center of the image are also defined, and they are two column matrices. Now I want to draw lines from the center to the boundary points so that I can obtain all points of intersection between these lines and the boundary of the image. How can I do that? Here is the code I have so far:
The code that is written just to get the one intersection point if anyone can help please
clear all
close all
clc
BW = im2bw(imread('C:\fyc-90_1-100.png'));
BW = imfill(BW,'holes');
[Bw m n]=preprocess(BW);
[bord sk pr_sk]=border_skeleton(BW);
boundry=bord;
L = bwlabel(BW);
s = regionprops(L, 'centroid');
centroids = cat(1, s.Centroid);
Step #1 - Generating your line
The first thing you need to do is figure out how to draw your line. To make this simple, let's assume that the centre of the human body is stored as an array of cen = [x1 y1] as you have said. Now, supposing you click anywhere in your image, you get another point linePt = [x2 y2]. Let's assume that both the x and y co-ordinates are the horizontal and vertical components respectively. We can find the slope and intercept of this line, then create points between these two points parameterized by the slope and intercept to generate your line points. One thing I will point out is that if we draw a slope with a vertical line, by definition the slope would be infinity. As such, we need to place in a check to see if we have this situation. If we do, we assume that all of the x points are the same, while y varies. Once you have your slope and intercept, simply create points in between the line. You'll have to choose how many points you want along this line yourself as I have no idea about the resolution of your image, nor how big you want the line to be. We will then store this into a variable called linePoints where the first column consists of x values and the second column consists of y values. In other words:
In other words, do this:
%// Define number of points
numPoints = 1000;
%// Recall the equation of the line: y = mx + b, m = (y2-y1)/(x2-x1)
if abs(cen(1) - linePt(1)) < 0.00001 %// If x points are close
yPts = linspace(cen(2), linePt(2), numPoints); %// y points are the ones that vary
xPts = cen(1)*ones(numPoints, 1); %//Make x points the same to make vertical line
else %// Normal case
slp = (cen(2) - linePt(2)) / cen(1) - linePt(1)); %// Solve for slope (m)
icept = cen(2) - slp*cen(1); %// Solve for intercept (b)
xPts = linspace(cen(1), linePt(1), numPoints); %// Vary the x points
yPts = slp*xPts + icept; %// Solve for the y points
end
linePoints = [xPts(:) yPts(:)]; %// Create point matrix
Step #2 - Finding points of intersection
Supposing you have a 2D array of points [x y] where x denotes the horizontal co-ordinates and y denotes the vertical co-ordinates of your line. We can simply find the distance between all of these points in your boundary with all of your points on the line. Should any of the points be under a certain threshold (like 0.0001 for example), then this indicates an intersection. Note that due to the crux of floating point data, we can't check to see if the distance is 0 due to the step size in between each discrete point in your data.
I'm also going to assume border_skeleton returns points of the same format. This method works without specifying what the centroid is. As such, I don't need to use the centroids in the method I'm proposing. Also, I'm going to assume that your line points are stored in a matrix called linePoints that is of the same type that I just talked about.
In other words, do this:
numBoundaryPoints = size(boundry, 1); %// boundary is misspelled in your code BTW
ptsIntersect = []; %// Store points of intersection here
for idx = 1 : numBoundaryPoints %// For each boundary point...
%//Obtain the i'th boundary point
pt = boundry(:,idx);
%//Get distances - This computes the Euclidean distance
%//between the i'th boundary point and all points along your line
dists = sqrt(sum(bsxfun(#minus, linePoints, pt).^2, 2));
%//Figure out which points intersect and store
ptsIntersect = [ptsIntersect; linePoints(dists < 0.0001, :)];
end
In the end, ptsIntersect will store all of the points along the boundary that intersect with this line. Take note that I have made a lot of assumptions here because you haven't (or seem reluctant to) give any more details than what you've specified in your comments.
Good luck.
I was wondering if you could advise me how I can connect several points together exactly one after each other.
Assume:
data =
x y
------------------
591.2990 532.5188
597.8405 558.6672
600.0210 542.3244
606.5624 566.2938
612.0136 546.6825
616.3746 570.6519
617.4648 580.4575
619.6453 600.0688
629.4575 557.5777
630.5477 584.8156
630.5477 618.5906
639.2696 604.4269
643.6306 638.2019
646.9013 620.7697
652.3525 601.1584
"data" is coordinate of points.
Now, I would like to connect(plot) first point(1st array) to second point, second point to third point and so on.
Please mind that plot(data(:,1),data(:,2)) will give me the same result. However, I am looking for a loop which connect (plot) each pair of point per each loop.
For example:
data1=data;
figure
scatter(X,Y,'.')
hold on
for i=1:size(data,1)
[Liaa,Locbb] = ismember(data(i,:),data1,'rows');
data1(Locbb,:)=[];
[n,d] = knnsearch(data1,data(i,:),'k',1);
x=[data(i,1) data1(n,1)];
y=[data(i,2) data1(n,2)];
plot(x,y);
end
hold off
Although, the proposed loop looks fine, I want a kind of plot which each point connect to maximum 2 other points (as I said like plot(x,y))
Any help would be greatly appreciated!
Thanks for all of your helps, finally a solution is found:
n=1;
pt1=[data(n,1), data(n,2)];
figure
scatter(data(:,1),data(:,2))
hold on
for i=1:size(data,1)
if isempty(pt1)~=1
[Liaa,Locbb] = ismember(pt1(:)',data,'rows');
if Locbb~=0
data(Locbb,:)=[];
[n,d] = knnsearch(data,pt1(:)','k',1);
x=[pt1(1,1) data(n,1)];
y=[pt1(1,2) data(n,2)];
pt1=[data(n,1), data(n,2)];
plot(x,y);
end
end
end
hold off
BTW it is possible to delete the last longest line as it is not related to the question, if someone need it please let me know.
You don't need to use a loop at all. You can use interp1. Specify your x and y data points as control points. After, you can specify a finer set of points from the first x value to the last x value. You can specify a linear spline as this is what you want to accomplish if the behaviour you want is the same as plot. Assuming that data is a 2D matrix as you have shown above, without further ado:
%// Get the minimum and maximum x-values
xMin = min(data(:,1));
xMax = max(data(:,1));
N = 3000; % // Specify total number of points
%// Create an array of N points that linearly span from xMin to xMax
%// Make N larger for finer resolution
xPoints = linspace(xMin, xMax, N);
%//Use the data matrix as control points, then xPoints are the values
%//along the x-axis that will help us draw our lines. yPoints will be
%//the output on the y-axis
yPoints = interp1(data(:,1), data(:,2), xPoints, 'linear');
%// Plot the control points as well as the interpolated points
plot(data(:,1), data(:,2), 'rx', 'MarkerSize', 12);
hold on;
plot(xPoints, yPoints, 'b.');
Warning: You have two x values that map to 630.5477 but produce different y values. If you use interp1, this will give you an error, which is why I had to slightly perturb one of the values by a small amount to get this to work. This should hopefully not be the case when you start using your own data. This is the plot I get:
You'll see that there is a huge gap between those two points I talked about. This is the only limitation to interp1 as it assumes that the x values are strictly monotonically increasing. As such, you can't have the same two points in your set of x values.
I am having difficulty with calculating 2D area of contours produced from a Kernel Density Estimation (KDE) in Matlab. I have three variables:
X and Y = meshgrid which variable 'density' is computed over (256x256)
density = density computed from the KDE (256x256)
I run the code
contour(X,Y,density,10)
This produces the plot that is attached. For each of the 10 contour levels I would like to calculate the area. I have done this in some other platforms such as R but am having trouble figuring out the correct method / syntax in Matlab.
C = contourc(density)
I believe the above line would store all of the values of the contours allowing me to calculate the areas but I do not fully understand how these values are stored nor how to get them properly.
This little script will help you. Its general for contour. Probably working for contour3 and contourf as well, with adjustments of course.
[X,Y,Z] = peaks; %example data
% specify certain levels
clevels = [1 2 3];
C = contour(X,Y,Z,clevels);
xdata = C(1,:); %not really useful, in most cases delimters are not clear
ydata = C(2,:); %therefore further steps to determine the actual curves:
%find curves
n(1) = 1; %n: indices where the certain curves start
d(1) = ydata(1); %d: distance to the next index
ii = 1;
while true
n(ii+1) = n(ii)+d(ii)+1; %calculate index of next startpoint
if n(ii+1) > numel(xdata) %breaking condition
n(end) = []; %delete breaking point
break
end
d(ii+1) = ydata(n(ii+1)); %get next distance
ii = ii+1;
end
%which contourlevel to calculate?
value = 2; %must be member of clevels
sel = find(ismember(xdata(n),value));
idx = n(sel); %indices belonging to choice
L = ydata( n(sel) ); %length of curve array
% calculate area and plot all contours of the same level
for ii = 1:numel(idx)
x{ii} = xdata(idx(ii)+1:idx(ii)+L(ii));
y{ii} = ydata(idx(ii)+1:idx(ii)+L(ii));
figure(ii)
patch(x{ii},y{ii},'red'); %just for displaying purposes
%partial areas of all contours of the same plot
areas(ii) = polyarea(x{ii},y{ii});
end
% calculate total area of all contours of same level
totalarea = sum(areas)
Example: peaks (by Matlab)
Level value=2 are the green contours, the first loop gets all contour lines and the second loop calculates the area of all green polygons. Finally sum it up.
If you want to get all total areas of all levels I'd rather write some little functions, than using another loop. You could also consider, to plot just the level you want for each calculation. This way the contourmatrix would be much easier and you could simplify the process. If you don't have multiple shapes, I'd just specify the level with a scalar and use contour to get C for only this level, delete the first value of xdata and ydata and directly calculate the area with polyarea
Here is a similar question I posted regarding the usage of Matlab contour(...) function.
The main ideas is to properly manipulate the return variable. In your example
c = contour(X,Y,density,10)
the variable c can be returned and used for any calculation over the isolines, including area.
My question is related to this link stackoverflow ques
In essence repeating the figure drawn there .. I have a central point ( x , y ) in an image around which I have to draw 4 circles of 1-4 unit radius with 8 angles between them.
In this diagram there are 12 angle bins but I have 8. There is a code solution there but it is for plotting the above figure.
I want to calculate the maximum intensity point in each of the 4 regions of each wedge. Is there any inbuilt function in matlab ? I looked at rose but could'nt understand if it would help me....
I would greatly appreciate if someone could help me how to calculate it in matlab....
Thanks
I put some code below that should be the basic skeleton of what you want to do. But I left an important function unimplemented because I think you will be able to do it and it will help you understand this process better.
% I assume that data_points is an M-by-2 array, where each row corresponds
% to an (x,y) coordinate pair, and M is the number of data points.
data_points = ... ;
% I assume this array stores the intensities at each data point.
intensities = ... ;
% I assume that this stores the total number of gridded polar regions you want
% to find the max intensity in (i.e. 4*(number of cells) in your picture above).
total_num_bins = ... ;
% This will store the max intensities. For places that have no nearby
% data points, the max intensity will remain zero.
max_intensities = zeros(total_num_bins);
% I assume these store the values of the center point.
x = ... ; y = ... ;
% The number of different data points.
num_data_points = length(intensities); % also equals size(data_points,1)
% Now, loop through the data points, decide which polar bin they fall in, and
% update the max intensity of that area if needed.
for ii = 1:num_data_points
% Grab the current point coordinates.
cur_x = data_points[ii,1];
cur_y = data_points[ii,2];
% Convert the current data point to polar coordinates,
% keeping in mind that we are treating (x,y) like the center.
cur_radius = sqrt( (cur_x - x)^2 + (cur_y - y)^2 );
cur_angle = atan2(cur_y - y, cur_x - x)
% You have to write this yourself, but it
% will return an index for the bin that this
% data point falls into, i.e. which of the 4 segments
% of one of the radial cells it falls into.
cur_bin = get_bin_number(cur_radius, cur_angle);
% Check if this data point intensity is larger than
% the current max value for its bin.
if ( intensities(ii) >= max_intensities(cur_bin))
max_intensities(cur_bin) = intensities(ii);
end
end
You will now have to make the function get_bin_number() which takes as its input the angle and radius of the data point away from the center point. It should return just an index between 1 and total_num_bins, because you will be keeping the max intensities in a linear array. So, for example, index number 1 might correspond to the first 1/4 piece of the closest radial cell in the upper right quadrant, index 2 might correspond to the next 1/4 of that same cell, moving counter-clockwise, or something like this. You have to devise your own convention for keeping track of the bins.
A late answer, but I believe an even easier solution would just be to convert your data from (x,y) coordinates to (r,theta) by using (r = sqrt(x.^2 + y.^2), theta = atan(y,x)) then use the hist3 function on the (r,theta) data set to get a radial histogram.
Therefore solution is as follows:
% I assume you have some M-by-2 matrix X that's in the form (x,y)
% Convert (x,y) to (r,theta)
xVect = X(:,1);
yVect = X(:,2);
X = [sqrt(xVect.^2 + yVect.^2), ...%formula for r
atan(yVect,xVect)]; %formula for theta
% 5 is the number of wedges along 'r', your radial axis
% 12 is the number of wedges along 'theta', your theta 'axis'
dist = hist3(X,5,12);
Even if you have solved this, I hope this helps anybody else who wants to create a radial/angular histogram!