Code covered by the BSD License  

Highlights from

4.1 | 27 ratings Rate this file 387 Downloads (last 30 days) File Size: 568 KB File ID: #7844 Version: 1.18
image thumbnail



David Legland (view profile)


13 Jun 2005 (Updated )

Geometry library for matlab. Performs geometric computations on points, lines, circles, polygons...

Editor's Notes:

This file was selected as MATLAB Central Pick of the Week

| Watch this File

File Information

Geometry library to handle and visualize geometric primitives such as points, lines, circles and ellipses, polylines and polygons...
The goal is to provide a low-level library for manipulating geometrical primitives, making easier the development of more complex geometric algorithms.
The library proposes functions to:
- create various shapes (points, circles, lines, ellipses, polylines, polygons) using an intuitive syntax. Ex: createCircle(p1, p2, p3) to create a circle through 3 points.

- derive new shapes: intersection between 2 lines, between a line and a circle, parallel and perpendicular lines
Functions to compute intersections

- work on polylines and polygons: compute centroid and area, expand, clip with half-plane...

- measure distances (between points, a point and a line, a point and a group of points), angle (of a line, between 3 points), or test geometry (point on a line, on a circle).

- manipulate planar transformation. Ex: P2 = transformPoint(P1, createRotation(CENTER, THETA));

- draw shapes easily. Ex: drawCircle([50 50], 25), drawLine([X0 Y0 DX DY]). Some clipping is performed for infinite shapes such as lines or rays.

Additional help is provided in geom/Contents.m file, as well as summary files like 'points2d.m' or 'lines2d.m'.

Note: the project has merged with the geom3d library (FeX 24484), and is now hosted on sourceforge:


Minimum Enclosing Circle and Curve Intersections inspired this file.

This file inspired Image Ellipsoid 3 D, Geometric And Topological Methods For Significance Testing In Wavelet Analysis, and Cumulative Areawise Testing In Wavelet Analysis.

Required Products Curve Fitting Toolbox
Image Processing Toolbox
Optimization Toolbox
Simulink Control Design
Statistics and Machine Learning Toolbox
Computer Vision System Toolbox
MATLAB release MATLAB 8.4 (R2014b)
MATLAB Search Path
Other requirements Some functions (for shape fitting) require the optimization toolbox.
Tags for This File   Please login to tag files.
Please login to add a comment or rating.
Comments and Ratings (76)
30 Nov 2015 Stratosphere


Is it possible to find intersection points on the polygon with equal angles to the center point? I mean that I want to choose a point N1 on the polygon and then find the following [N1,N2,...,N36] points on the polygon such that:

Angle between N1 and Center = Theta
Angle between N2 and Center = Theta + 10
Angle between N3 and Center = Theta + 20
Angle between N4 and Center = Theta + 30
Angle between N36 and Center = Theta + 350

Which functions can help me? Thanks.

Comment only
11 Nov 2015 Thayjes Srivas

I have a question on how exactly we could reorder the points of a polygon such that all the self intersections are removed ? I have tried playing around with the polygonSelfIntersection and polygonLoops function but still not figured out how to reorder the points. Thanks

Comment only
08 Oct 2015 filipp

filipp (view profile)

05 Oct 2015 David Legland

David Legland (view profile)

Hi Filipp,
the function "projPointOnPolygon" does not rely on any toolbox, so I do not understand the difference in speed. The underlying function is "distancePointEdge", that I tried to optimize using bsxfun. Do no work with large number of vertices? Have you tried to profile to identify where code is slow? (You can contact me by mail, this will be easier to exchange)
Regards, David

Comment only
02 Oct 2015 filipp

filipp (view profile)

Hi, I have a question on the projPointOnPolygon function. It is running extremely slow on one system with Matlab 2014b (no toolboxes installed), but works properly on another system with Matlab 2012 (most toolboxes installed). Is it using any toolboxes? Are there other reasons why it might be so slow (given that the first system even has more RAM)? Thank you!

Comment only
21 Sep 2015 Eric

Eric (view profile)

06 Aug 2015 Krisztian Szucher  
01 Jun 2015 David Legland

David Legland (view profile)

In the case you propose, the test line overlap one of the edges of the polygon. The function intersectLinePolygon does not handle such degenerate cases. You can avoid it by slightly shifting the test line for example.

Comment only
29 May 2015 Tao

Tao (view profile)

I probably met a bug in intersectLinePolygon.m

I call this function by:

line=[0 11.8760 1 0];
offsets= [-6.0960 11.8760;
-6.0960 11.6760;
-6.7870 11.6760;
-9.1180 11.8330;
-9.1180 11.8760;
-6.0960 11.8760];

then I get: intersect =

-6.0960 11.8760

But it is supposed be: intersect =

-9.1180 11.8760
-6.0960 11.8760

For other situations ,it works well except this one.

Can you work it out?? My MATLAB version is : R2012a

Comment only
13 May 2015 David Legland

David Legland (view profile)

I have updated a new version, with some updates on medialAxisConvex. It works on the few polygons I have tested. I hope it will fit your needs.

Comment only
03 May 2015 Jan Bartonicek

I tried to run demoMedialAxis and got this error:
??? Error using ==> vertcat
CAT arguments dimensions are not consistent.

Error in ==> medialAxisConvex at 123
events = [events; ir(1) i pint(1,:) ti(1); ir(2) i pint(2,:) ti(2)]; %#ok<AGROW>

There may be some problem with point adding. Do you know where?
Thank you

Comment only
07 Apr 2015 David Legland

David Legland (view profile)

@bn dhouba:
an edge in geom2d should be understood as 'line segment'. Is defined by two extremities, and is restricted to the set of points between these two points. The functions working on 'line' objects considers infinite straigt lines. Hope this makes things clearer? regards,

Comment only
05 Apr 2015 bh dhouha

Hi, i really apprciate this package it's really intresting. I just dont understand the exact meaning of edge here in the functions DrawEdge or intersecEdges , do you mean simply line? is it the same thing?

Comment only
02 Dec 2014 David Legland

David Legland (view profile)

Hi William,
functions meshVolume and polygonArea depends on the face (or vertices) orientation. If all faces are oriented outwards, the volume will be positive. If all facets are orieted inwards, you will obtain the negative value of the volume. Just be careful to ensure faces orientation is consistent (use checkMeshAdjacentFaces for example)

Comment only
01 Dec 2014 William

I seem to get a negative volume using the meshVolume function...
I'm trying to calculate the volume of a bendy open ended cylinder, using the faces and vertices generated from an isosurface patch. Any thoughts?

Comment only
27 Oct 2014 David Legland

David Legland (view profile)

Hi Jasper,
thanks for pointing out the bug! I have released a new version that should fix it. Tell me if there are still problems.

Comment only
23 Oct 2014 Jdeen

Jdeen (view profile)

I believe I found a bug in the clipEdge function, however do not have the time to investigate what exactly the bug is. However, results can be different when calling clipEdge(allEdges, box) instead of clipEdge(oneEdge, box) for each box. Example variables where the error occurs can be found here:
Edge in row number four is not correctly clipped when clipping all edges at once.

14 Aug 2014 Fadi Kahwash

This is a treasure for beginners in computational geometry.
Many Thanks,


27 Jun 2014 filipp

filipp (view profile)

Hi David,
thanks a lot for your suggestion. Using Delaunay triangulation sounds reasonable and I will definitely try that!

25 Jun 2014 David Legland

David Legland (view profile)

Hi Filipp,
I think your problem is closely related to Hamiltonian path in graph theory. Maybe what you can try to do is create Delaunay triangulation of your points. This will produce a set of triangle, that you can convert to a set of edges that do not self-intersect. The set of edges form a graph, then you can find or write an algorithm that finds an Hamiltonian path that goes through each vertex once and only once. Hope this gives the idea...

By "sorting" vertices, I was meaning finding the order of indices that need to be considered. Maybe the term "ordering" would have been more appropriate. The idea is to keep the original array unchanged, and work only on vertex indices. But this is just technical consideration.


Comment only
23 Jun 2014 filipp

filipp (view profile)

Hi David,
thanks for the patience :-) The second option would be the one that I am searching for: finding a path going through a set of (random) points without self-intersecting. I already tried a brute force method: choosing a random starting point, connecting points one-by-one, and going back when intersections are found. However, this is extremely costly and did not seem to be an elegant solution. So, what do you mean by sorting vertices of interest?

Comment only
20 Jun 2014 David Legland

David Legland (view profile)

Hi Filipp,
I am not sure to really understand your question... Do you want to draw the convex hull of a set of points? Or some kind of path going through a set of points without self-intersecting? In both cases, you first need to identify and/or sort vertices of interest before plotting them.

Comment only
17 Jun 2014 filipp

filipp (view profile)

Hi David,
thanks for your response. I've already tried these two functions, however, both allow for intersection. As a consequence, they do not result in a shape with a single circumferential contour but in a jumble of crossing lines. Do you have an alternative idea how to create closed contours from a number of random points?

Comment only
10 Jun 2014 David Legland

David Legland (view profile)

Hi filipp,
yes, you can use one of the "drawPolygon" or "fillPolygon" functions (in the "polygons" subpackage). Hope this helps ?


Comment only
28 May 2014 filipp

filipp (view profile)

Hi David,
I search for a way to display polygons without intersecting faces (i.e., shapes with one contour). Is this function available in your package or in matgeom? If not, do you have an idea to implement this?

Comment only
27 May 2014 Ralph Trimnell

Finally, works good! Now for the real fun! Thanks, again, Ralph

Comment only
26 May 2014 Ralph Trimnell

Thanks, David,
Maybe I'm out of my depth here but geom looked ideal to create simulations of organic growth events as an addition to the jdk. Will it not work without Matlab?

Comment only
26 May 2014 David Legland

David Legland (view profile)

Hi Ralph,
There is no need to download java for using this contribution. What you get after download is a zip file that contains a set of '.m' files, in a small directory arborescence, that should be run with Matlab.
Hope this makes things clearer?

Comment only
26 May 2014 Ralph Trimnell

Ralph Trimnell

Could someone explain how to get Java to recognize geom I tried pasting the whole JAR into the jdk:no good! I put it's address in the system environment path: no good. Once extracted, it's no longer a Jar file,just has extention geom. There should be an easy way to do this. I'm using notepad on windows 8.1. Thanks

Comment only
01 Apr 2014 David Legland

David Legland (view profile)

Hi Gary,

The "expansion" or "erosion" of the polygon depends on polygon orientation (counter-clockwise or clockwise) and on the sign of the distance. If polygon is CCW and distance is positive, the result will be expanded.

I have submitted a new version, that fixes some bugs. It seems to work better.


Comment only
31 Mar 2014 Gary Kenward

The function expandPolygon appears to have a number of problems.

The following shrinks the polygon rather then expand it:

Sq1 = [-10.28,-15.39;-17.28,-4.611;-11.89,-1.110;-4.890,-11.89]
Sq1 = [ Sq1; Sq1(1,:) ];
Sq1e = cell2mat ( expandPolygon ( Sq1, 1 ) );

And, the following generates an error within the function:

Sq = [ [-10, -5]; [-10, 5]; [-5, 5]; [ -5, -5] ];
Sq1 = (Sq * [ cos(33*pi/180) -sin(33*pi/180); sin(33*pi/180) cos(33*pi/180)]' ) * 9/7 - 3;
Sq1 = [ Sq1; Sq1(1,:) ];
Sq1e = cell2mat ( expandPolygon ( Sq1, 1 ) );

The error message is:

Attempted to access poly(0,:); index must be a
positive integer or logical.
Error in distancePointPolygon (line 19)
if sum(poly(end, :)==poly(1,:))~=2
Error in distancePolygons (line 19)
dist2 = min(distancePointPolygon(poly2, poly1));
Error in expandPolygon (line 63)
distLoop(i) = distancePolygons(loops{i}, poly);

16 Jan 2014 Boris Schauerte

Exactly what I was looking for. Well documented and easy to use. Thanks!

29 Oct 2013 J.R.! Menzinger

nice work.

08 Mar 2013 Marc

Marc (view profile)

Really nice works... Thanks

08 Mar 2012 David Legland

David Legland (view profile)

Hi sumitha,
basic format for representing lines is [x0 y0 dx dy], where (w0,y0) refers to origin point, and (dx,dy) to direction vector. You can create a line orthogonal to direction THETA with:
line1 = [x0 Y0 -sin(THETA) cos(THETA)];

Comment only
07 Mar 2012 sumitha

Hi,can u please suggest me which one of these i can use to draw a line orthogonal to the given orientation(in radian)or else can u plz give me an idea how i can proceed with the same.

01 Mar 2012 David Legland

David Legland (view profile)

Hi Benjamin, the clipping of polygon is not an easy problem. There are efficient algorithms, and their inclusion is planned, but I still haven't found time to write them down. It seems the mapping toolbox provides facilities for this, maybe you can check it.
Regards, David

Comment only
28 Feb 2012 Benjamin Sanderse

Hi David, thanks for the great work. I have been using your clipPolygon successfully for clipping convex polygons, but now I also need it for non-convex polygons... are you planning to include this in a future release?

18 Feb 2012 SHAFIQUL

Very nice and essential coding thanks for sharing.

Comment only
19 Nov 2011 marriam nayyer

anyone please help me

Comment only
06 Jun 2011 David Legland

David Legland (view profile)

@Frank: yes, there was a stupid bug in this function. I have submitted an enhanced version that fixes it.
Regards, David

Comment only
02 Jun 2011 Frank Gommer

Hi, looks like a nice little toolbox!
I just tried your function "intersectLineCircle.m". Unfortunately, it seems that the code does not give the right answers for tangential lines (Returns 'nan') and lines which are not intersecting at all (gives a wrong intersection point).

Here is the data I used:
c1 = [0 0 0.5]
l1 = [0.5 0 0 1] %tangential
l1 = [-1 0 0 1] %non touching
pts = intersectLineCircle(l1, c1)


Comment only
08 Apr 2011 Juan Pablo Carbajal

geom2d is a first class complement to the geometry tools provided.
David: You should add a 2D alpha shape routine. I am testing this one at the moment, but leaves lots of room for improvement.

16 Dec 2010 David Legland

David Legland (view profile)

Hi Chris,
intersection of circle with line/circle is not implemented for the moment, juste because I did not need it yet... But I will try to implement it in a future release.

Comment only
11 Dec 2010 Chris TIAN

Hi, this is very useful for me? I just wonder why didn't i find the function for obtaining intersections between a line and a circle? Thanks.

Comment only
01 Dec 2010 Ben

Ben (view profile)

Well done, David!

14 Sep 2010 Reto Zingg

OK, I think this should work, circumventing eps. Again, looking at center points instead of end-points. This time the center points of segments. It still needed some fail-safe code when multiple edge segments show up inside box due to finite precission. The code should also maintain 'direction' of the edges, as your previous code.
Here the conplete code for clipEdge:

% process data input
if size(box, 1)==1
box = box([1 3 2 4]);

% get limits of window
xmin = box(1);
ymin = box(2);
xmax = box(3);
ymax = box(4);

% convert window limits into lines
lineX0 = [xmin ymin xmax-xmin 0];
lineX1 = [xmin ymax xmax-xmin 0];
lineY0 = [xmin ymin 0 ymax-ymin];
lineY1 = [xmax ymin 0 ymax-ymin];

edge2 = zeros(size(edge));

for i=1:size(edge,1)
% current edge
iedge = edge(i, :);

% compute intersection points with each line of bounding window
px0 = intersectLineEdge(lineX0, iedge);
px1 = intersectLineEdge(lineX1, iedge);
py0 = intersectLineEdge(lineY0, iedge);
py1 = intersectLineEdge(lineY1, iedge);

% create array of points
points = [px0; px1; py0; py1; iedge(1:2); iedge(3:4)];

% and remove infinite points (edges parallel to box edges)
points = points(all(isfinite(points),2), :);

% sort points by x then y
points = sortrows(points);

% get center positions between points

% find the centerPoints (if any) inside window
ins = find( centerPoints(:,1)>=xmin & centerPoints(:,2)>=ymin & ...
centerPoints(:,1)<=xmax & centerPoints(:,2)<=ymax);

% if multiple segments inside box, which can happen due to finite resolution, only take the longest segment
if length(ins)>1
dv=points(ins+1,:)-points(ins,:); % delta vectors of the segments
len=hypot(dv(:,1), dv(:,2)); % lengths of segments
[a,I]=max(len); % find index of longest segment

if length(ins)==1 % if one of the center points is inside box, then the according edge segment is indide box
if edge(i,1)>edge(i,3) || (edge(i,1)==edge(i,3) && edge(i,2)>edge(i,4)) % restore same direction of edge
edge2(i, :) = [points(ins+1,:) points(ins,:)];
edge2(i, :) = [points(ins,:) points(ins+1,:)];

Comment only
14 Sep 2010 Reto Zingg

Argh, never mind. My solution has div by 0 problem. The whole eps stuff is really a pain in the ...

Comment only
14 Sep 2010 Reto Zingg

turns out my suggestion above introduces a problem with negative numbers. The code below should work correctly:

% ins = find( points(:,1)-xmin>=-eps & points(:,2)-ymin>=-eps & ...
% points(:,1)-xmax<=eps & points(:,2)-ymax<=eps);
ins = find( (points(:,1)/xmin-1)*sign(xmin)>=-eps & (points(:,2)/ymin-1)*sign(ymin)>=-eps & ...
(points(:,1)/xmax-1)*sign(xmax)<=eps & (points(:,2)/ymax-1)*sign(ymax)<=eps);

Comment only
13 Sep 2010 Reto Zingg

Also in clipEdge() you should change the limit check. See code below. Change commented out to suggested.


% ins = find( points(:,1)-xmin>=-eps & points(:,2)-ymin>=-eps & ...
% points(:,1)-xmax<=eps & points(:,2)-ymax<=eps);
ins = find( points(:,1)/xmin-1>=-eps & points(:,2)/ymin-1>=-eps & ...
points(:,1)/xmax-1<=eps & points(:,2)/ymax-1<=eps);

Comment only
09 Sep 2010 Reto Zingg

again some trouble with large numbers and limit checking. This time in isPointOnEdge, which is used for intersectLineEdge, which in turn is used in clipEdges.
isPointOnEdge will not work correctly for coordinates with large numbers, as you apply a fixed tolerance. However, the numerical tolerance of matlab is relative to the number itself.
I normalized the point and edge x and y elements to the point x and y respectively. Like that you will deal with numbers in the range of 1 and your tolerance check will work.
See the code below. Note that I also dumped the Np==Ne check, as it didn't do anything.
Regards, Reto

% extract computation tolerance
tol = 1e-14;
if ~isempty(varargin)
tol = varargin{1};

% number of edges and of points
Np = size(point, 1);
Ne = size(edge, 1);

% adapt size of inputs
x0 = repmat(edge(:,1)', Np, 1);
y0 = repmat(edge(:,2)', Np, 1);
dx = repmat(edge(:,3)', Np, 1)-x0;
dy = repmat(edge(:,4)', Np, 1)-y0;
xp = repmat(point(:,1), 1, Ne);
yp = repmat(point(:,2), 1, Ne);

% normalize x, y to xp, yp
x0 = x0/xp;
y0 = y0/yp;
dx = dx/xp;
dy = dy/yp;
xp = 1;
yp = 1;

% test if point is located on supporting line
b1 = abs((xp-x0).*dy - (yp-y0).*dx)./hypot(dx, dy)<tol;

% check if point is located between edge bounds
% use different tests depending on line angle
ind = abs(dx)>abs(dy);
t = zeros(max(Np, Ne), 1);
t(ind) = (xp(ind)-x0(ind))./dx(ind);
t(~ind) = (yp(~ind)-y0(~ind))./dy(~ind);
b = t>-tol & t-1<tol & b1;

Comment only
05 Aug 2010 David Legland

David Legland (view profile)

this is a good idea, much more convenient than dealing with eps. I will implement it soon.
thanks for advices !

Comment only
04 Aug 2010 Reto Zingg

Hi David,
your new limit check assumes that the direction vector of the line has a similar scale as the bounding box, which is not necessarely true. The direction vector often has a magnitude of 1, while the bounding box can be numerically very large.

Here is an idea and the thinking behind it:
You are checking the endpoints of the edge to be within the bounding box. Naturally, the end points are always on the bounding box, so you run into numerical precission issues. So why not check if the center of the edge is within the box? This way you are away from the limit lines with the exception of horizontal and vertical lines. In those two cases you should not have any numerical issues, as they are at right angles.

So, I would re-write the limit check as follows and drop eps completely:

% check if edge center is within bounds
xMinOk = xmin <= edgeCenterX;
xMaxOk = edgeCenterX <= xmax;
yMinOk = ymin <= edgeCenterY;
yMaxOk = edgeCenterY <= ymax;

It works for me.

Comment only
04 Aug 2010 David Legland

David Legland (view profile)

Hi Reto,

thank you for your numerous and constructive remarks. I have published a new version that take into account your corrections. I will try to propagate modification to other drawing functions.

For limit check, I have used a different strategy, but it seems to work with large bouding boxes as well. Tell me if there are still problems.

Comment only
03 Aug 2010 Reto Zingg

Also consider re-writing adding the precision to the limit check as follows:

xOk = xmin*(1-eps)-min(edge(:,[1,3]), [], 2)<0 & max(edge(:,[1,3]), [], 2)-xmax*(1+eps)<0;
yOk = ymin*(1-eps)-min(edge(:,[2,4]), [], 2)<0 & max(edge(:,[2,4]), [], 2)-ymax*(1+eps)<0;

This will take into consideration the scale of the limiting box. With the original code you run into precission issues when the box has large numbers (I'm working with numbers in the 1e9 range). Like this the precision is relative rather than absolute.

Comment only
03 Aug 2010 Reto Zingg

Yet an other bug. This library is a good start, but definitely not ready fro prime time.
in clipLine() there is incorrect indexing. Fix is below. Consider using this fixed function in drawLine() (see bug in drawLine() above).

% check bounds of result edge in each direction

% incorrect indices, indexing a edge, not a box
% xOk = xmin-min(edge(:,1:2), [], 2)<eps & max(edge(:,1:2), [],2)-xmax<eps;
% yOk = ymin-min(edge(:,3:4), [], 2)<eps & max(edge(:,3:4), [],2)-ymax<eps;

% this is correct indexing
xOk = xmin-min(edge(:,[1,3]), [], 2)<eps & max(edge(:,[1,3]), [], 2)-xmax<eps;
yOk = ymin-min(edge(:,[2,4]), [], 2)<eps & max(edge(:,[2,4]), [], 2)-ymax<eps;

02 Aug 2010 Reto Zingg

oops, now my correction of the correction replaced the original correction. You can tell I'm a novice to this system...
So, here again the complete suggested correction to drawLine().
The original code will draw horizontal lines even though they are outside the axis. It will also draw lines in the quadrant to the right and below the axis.

% sort points along the x coordinate, and draw a line between
% the two in the middle
points = sortrows([px1 ; px2 ; py1 ; py2], 1);

if isfinite(points(3,1)) % if not horizontal or vertical
else % else horizontal or vertical

if points(1,1)>=xmin && points(2,1)<=xmax &&...
points(1,2)>=ymin && points(2,2)<=ymax

h(i)=line(points(1:2, 1), points(1:2, 2));

if ~isempty(varargin)
set(h(i), varargin{:});


Comment only
02 Aug 2010 Reto Zingg

Sorry, correction to the correction:

if points(1,1)>=xmin && points(2,1)<=xmax &&...
points(1,2)>=ymin && points(2,2)<=ymax

Comment only
02 Aug 2010 Reto Zingg

There is a bug in intersectEdges(). Change lines 143 and 144 from

x0(out) = NaN;
x0(out) = NaN;


x0(i(out)) = NaN;
y0(i(out)) = NaN;

otherwise indexing is incorrect and you get false results.

20 Jul 2010 David Legland

David Legland (view profile)

I have submitted a new version, that fixes the bug in the 'centroid' function, as well as many others.

Comment only
13 Jul 2010 Matthew

Minor (but fatal) bug found in centroid.m for calculating mass-weighted centroid of 2d object.

Change line 69 of the file from:

center = mean(pts.*repmat(mass, 1, 3));


center = mean(pts.*repmat(mass, 1, 2));

Or it will error with a subscript assignment dimension mismatch...pts is a 2-column matrix, not a 3-column matrix.

03 Jun 2010 Juan Pablo Carbajal  
19 May 2010 David Legland

David Legland (view profile)

Hi Aleksandar,
at the moment there is no additional argument for drawRect function.
To specify the figure to draw in, the best is to use figure(...) before calling drawRect, eventually using hold on to avoid clearing previous image.
To change rectangle color or width, you can use the handle returned by the function:
hr = drawRect([x0 y0 w h]);
set(hr, 'color', 'r', 'linewidth', 4);

Comment only
17 May 2010 Aleksandar

I need help with function drawRect.
Can this function have more arguments? (Like: image on which to put rectangle,rectangle color ...)

I have this part of code
picture = double(imread('test2.jpg'));
im = drawRect(...
round((vj(pos, 2) - 1) * sF + 1),...
round((vj(pos, 2) + velicinaLica(1) - 1) * sF + 1),...
round((vj(pos, 3) - 1) * sF + 1),...
round((vj(pos, 3) + velicinaLica(2) - 1) * sF + 1),...
[1 0 0]...

Comment only
24 Jul 2009 David Legland

David Legland (view profile)

oops, sorry for badly corrected bug...
A new version has been submitted, with suggested corrections. There are also some new functions for polygons and polylines (see for example "help polygons2d" or "help polylines2d")

Comment only
23 Jul 2009 Bala Krishnamoorthy

I tested some of the other functions. Quite useful indeed! Thanks.

22 Jul 2009 Bala Krishnamoorthy

The file intersectLinePolygon.m still contains 'pi' in place of 'intersects'. The HISTORY says it has been replaced, but it has NOT been replaced yet. I downloaded the zip file on July 22, 2009.

Also, here is a clearer (and of course equivalent way) to do the update:

intersects = [];
for i=1:length(edges)
p0 = intersectLines(line, createLine(edges(i,1:2), edges(i,3:4)));
if isPointOnEdge(p0, edges(i,:))
intersects = [intersects; p0];

Comment only
17 Jul 2009 Nathan Thern

This library and its companion, geom3d, are shining jewels - indispensable for doing non-trivial calculations on objects in 2d and 3d space.

By the way, leave all the files in the geom2d directory and add the directory to your path. Then you can type "help geom2d" or "doc geom2d" and get properly linked help text in the command window or the help window. The Contents.m file is the proper "MATLAB way" of providing documentation.

18 Jun 2009 David Legland

David Legland (view profile)


I've submitted a new version, which fixes the bug in intersectEdges, and adds new functions for manipulating polylines (intersections, distance to point...). Please do not hesitate to report bugs (direct email preferred).

The 'pi' variable (cf. alexia's comment) was not supposed to cause problem, as it is located inside of a function. The goal of the library is mainly to hide computational details into functions with explicit names, without having to read the inner code. However, the variable 'pi' was renamed in a more appropriate (and hopefully explicit) name.

regards, David

Comment only
11 Jun 2009 Greg

Greg (view profile)

I've tried the function intersectEdges but found that some cases are mishanlded etc.

23 Nov 2008 alexia lavadens

Hi, what a bad name for a variable/function "pi", I mean:

"pi=intersectLinePolygon(line, points)"

I had a headache trying to discover what was wrong with this file (because in theory it should work), it was part of a very large project I was working in.

Finally I discovered the "pi" issue, "pi" is present in the code, but I also had to work with pi=3.1415..., and this cr*p of matlab got confused. Oh Mathematica !, I have such good memories...

Certainly is not so clever call a variable/function "pi" in a geometry script. Besides, as peter said is very badly commented (found errors), so don't trust to much with this one.

23 Nov 2008 alexia lavadens  
06 Jun 2007 Nilimb Misal

very helpful in implementing more complicated geometry algorithms

08 Mar 2007 Matthew Davidson

drawLabels doesn't actually draw any labels...

07 Nov 2005 Peter Gurk

Tried circleaspolygon and rectaspolygon but both were not self contained, badly explained. The package seems to be promising but far from plug and play.

14 Jun 2005 Stephen McNeill

Very handy functions for geometric analysis

12 Apr 2006

update doc, add some functions, fix some bugs

06 Jun 2006

update description

05 Nov 2010 1.11

bug fixes (isPointOnEdge, intersectPolylines), remove obsolete licence info

05 Nov 2010 1.12

fix bugs (intersectLines, isPointOnEdge), remove obsolete licence considerations

30 Mar 2011 1.13

split library in several packages, add several functions (circle and circle-line intersections, computation of inertia ellipse)

06 Jun 2011 1.14

fixed bugs in intersectLineCircle, added minimumCaliperDiameter and averagePointSet of a set of points. Now uses degrees for representing shapes

29 Feb 2012 1.15

lot of code cleanup, new functions for working with polygons, clean up drawing functions

01 Apr 2014 1.16

various code cleanup and doc updates

27 Oct 2014 1.17

fix several bugs (clipEdge, removeMultipleVertices...), various doc update, add some demos for polynomial curves

13 May 2015 1.18

update medialAxisConvex, add smoothPolygon, smoothPolyline, findClosestPoint functions

Contact us