• FEATURED

View more

View more

View more

### Image of the Day Submit

IOTD | Top Screenshots

### The latest, straight to your Inbox.

Subscribe to GameDev.net Direct to receive the latest updates and exclusive content.

# Vectorizing a 1-bit raster image of polygon(s)

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

14 replies to this topic

### #1CGameProgrammer  Members

Posted 07 January 2014 - 07:23 PM

I have a one-bit image (each pixel is either black or white) of a polygon or polygons - the edges of the image are always black. I would like to convert this to a set of one or more polygon vectors, where each vector is simply an ordered list of points (one list per polygon) describing the corners or approximating curves.

Attached is an example input image with red dots overlayed to show what the output points should be. It should be two lists of points of course. The threshold for placing points around a curve is arbitrary and not important.

The input polygons will never have holes.

Speed is irrelevant; this is not being done at realtime and is just precomputed.

The actual scenario, if you're wondering, is that I have airspace regions defined as simple polygonal regions (one list of coordinates per polygon) which are either unioned or subtracted. The sample image illustrates an example of this (not actual airspace, just a simple test). I started with a hexagon, then subtracted from it a quadrilateral (which splits the hexagon in two), then I added (unioned) a circle. Trying to directly generate vector polygons from these input polygon vectors would be maddening; I figure it's simpler to render them as raster polygons temporarily (additive is drawn in white, subtractive in black) and then convert that.

Step one, edge detection, becomes trivial obviously; it's any white pixel adjacent to a black pixel.

Step two is to create huge unsimplified ordered lists of edge points. My plan is to create a bucket of all edge pixels, then (sloppy pseudocode):

point = bucket.pop()

isDone = false

while !isDone
{
isDone = true

for all edge pixels in bucket
{
if this edge pixel is adjacent to point
{
append edge pixel to polygon points
point = edge pixel
isDone = false
break;
}
}
}


Then if there is anything remaining in the bucket, repeat the process for the next new polygon.

The third and final step, and one that I definitely can't seem to figure out, is to simplify these lists of every edge into just the ones that mark the corners (or that approximate to some threshold the curves).

Can anyone help?

#### Attached Thumbnails

Edited by CGameProgrammer, 07 January 2014 - 07:47 PM.

~CGameProgrammer( );
Developer Image Exchange -- New Features: Upload screenshots of your games (size is unlimited) and upload the game itself (up to 10MB). Free. No registration needed.

### #2Erik Rufelt  Members

Posted 07 January 2014 - 07:36 PM

Trying to directly generate vector polygons from these would be maddening; I figure it's simpler to render them as raster polygons temporarily (additive is drawn in white, subtractive in black) and then convert that.

I disagree, this should be simpler, though it depends on what you want to use it for. What is the purpose of having a representation as a concave polygon?

If you need to check intersections or similar with the resulting shapes, then I would not precompute a final polygon shapes at all, but rather use the original simple polygons, including subtractions. This is not terribly difficult for boolean operations on constructive solid geometry (CSG). For example checking if a point is inside a tree of polygons combined through boolean operations can be easily reduced to checking only if a point is inside the simple sub-polygons and then inverting the answer depending on subtraction. You can even do subtractions of subtractions with ease, for example subtracting a rectangle from a circle, then subtracting the resulting shape from another circle, etc.

If you wish to go backwards from rasterization, I would probably use a flood-fill algorithm to find each complete white shape and then trace the edge of each such shape. Go through every pixel until you find a white one, flood-fill around it to find every white pixel that is connected to that pixel, and you have one shape. Then trace around it. Then remove all points that are exactly on a straight line between two other points to reduce the number of edge-points to corners.

Edited by Erik Rufelt, 07 January 2014 - 07:41 PM.

### #3CGameProgrammer  Members

Posted 07 January 2014 - 07:44 PM

I know it's very easy to test if something is within the original simple polygons; my problem is rendering them. There is no way to directly render the original polygons that would look correct. Now it is trivial to render the raster polygons with a border; I'd just draw a circle (diameter = border thickness) at each edge pixel. The problem with that is of course it'll look blurry when stretched and pixellated when shrunk. I'd be rendering them on a map so the user would zoom a lot.

Also since we're dealing with pixels, they are not on a straight line due to interpolation. For example:

X
XX
X
XX


Edited by CGameProgrammer, 07 January 2014 - 07:45 PM.

~CGameProgrammer( );
Developer Image Exchange -- New Features: Upload screenshots of your games (size is unlimited) and upload the game itself (up to 10MB). Free. No registration needed.

### #4Erik Rufelt  Members

Posted 07 January 2014 - 08:24 PM

I see. I would do CSG of the polygons. Triangulate each simple polygon for rendering, then for the subtractions you can reduce the problem to triangle/triangle subtraction, which is rather simple, especially when you don't care too much about limiting the number of triangles, and you get rid of any interpolation problems.

### #5CGameProgrammer  Members

Posted 07 January 2014 - 09:17 PM

No, that would not work because I am not rendering a triangular mesh; the output must be the list of points that define the polygon's border.
~CGameProgrammer( );
Developer Image Exchange -- New Features: Upload screenshots of your games (size is unlimited) and upload the game itself (up to 10MB). Free. No registration needed.

### #6Erik Rufelt  Members

Posted 07 January 2014 - 09:39 PM

Explain how the final render should look and I'm sure there is a relatively straight-forward way to accomplish it.

### #7CGameProgrammer  Members

Posted 08 January 2014 - 12:30 AM

I appreciate you trying to help but you keep suggesting inappropriate alternatives instead of the solution to the question I asked. In this case the polygon is added to Google Maps as a list of coordinates so it can be properly vector-drawn; adding it as a raster image would look much worse as I mentioned in the original post.

If it wasn't obvious, and maybe it wasn't, the answer I really need is the solution to step #3 in the original post: reducing an ordered list of points along the edges to just the corners (except for curves).
~CGameProgrammer( );
Developer Image Exchange -- New Features: Upload screenshots of your games (size is unlimited) and upload the game itself (up to 10MB). Free. No registration needed.

### #8Krypt0n  Members

Posted 08 January 2014 - 06:05 AM

the simplest way I can think of is

1. calculate for every vertex and line that you could 'remove' what error it would cause

2. remove the vertex or line with the lowest error.

3. goto 1

calculating the error:

assume you have

struct line
{
int IndexVertex0;
int IndexVertex1;
std::vector<int> RemovedVertices;
};
struct Shape
{
std::vector<vec2> Vertices;
std::vector<line> Lines;
};


in a stupidly simple way:

cheapest=invalid;
minerror=infinity;
for_each vertex
Shape Tmp=SourceShape
Tmp.remove(vertex)
error=calculate_cost(Tmp);
if(error<minerror)...

for_each line
Shape Tmp=SourceShape
Tmp.remove(line)
error=calculate_cost(Tmp);
if(error<minerror)...

if(minError>maxAllowedError) return;

SourceShape.remove(...)


"remove"

not only removes a vertex or a line, BUT

-the removed vertex is added to line.RemovedVertices

-the removed line (which also removes two vertices!) is added to line. Remove

... of the new line that spans in place of the old vertex or line

"calculate_cost"

accumulates the cost for every line in the shape

the cost of every line is the sum of squared shortest distances of all vertices in "RemovedVertices"

error=0;
for_each vertex in Line.RemovedVertices
Dist=vertex.ShortestDistanceTo(Line)
Error+=Dist*Dist


that's similar to quadric error metric in mesh simplification, yet with your dicritized source shape, you have a very blocky source and you want to end up with a smoother 'better' shape, that will lead to an error that will usually increase at first, but once the lines get longer and fit better the real lines, the error will decrease again. that's why you need to calculate the error per line again.

hope that helps

### #9Erik Rufelt  Members

Posted 08 January 2014 - 07:44 AM

I appreciate you trying to help but you keep suggesting inappropriate alternatives instead of the solution to the question I asked. In this case the polygon is added to Google Maps as a list of coordinates so it can be properly vector-drawn; adding it as a raster image would look much worse as I mentioned in the original post.

If it wasn't obvious, and maybe it wasn't, the answer I really need is the solution to step #3 in the original post: reducing an ordered list of points along the edges to just the corners (except for curves).

I assure you that skipping rasterization is not an inappropriate solution. But obviously there is more than one way to do things, and apologies if I misunderstood anything and good luck

### #10CGameProgrammer  Members

Posted 08 January 2014 - 11:45 AM

Your idea may have some merit Krypt0n. I'm not positive that calculating length is the right way to determine error; I worry it may either cut vital corners if it's too aggressive or fail to remove enough points on curves if it's not aggressive enough. Or both. But I can try it and see.

Edited by CGameProgrammer, 08 January 2014 - 11:46 AM.

~CGameProgrammer( );
Developer Image Exchange -- New Features: Upload screenshots of your games (size is unlimited) and upload the game itself (up to 10MB). Free. No registration needed.

### #11Krypt0n  Members

Posted 08 January 2014 - 12:19 PM

imagine two lines meeting at a vital corner. those lines have a lot of point in "RemovedVertices", even slight changes in direction accumulate to a big error. removing a vital corner -> very big error. on the other side, removing vertices on the path between two vital corners won't affect the minimal distance between every removed vertex and the line.

that was of course just the simpelst of the ways to calculate the error. you can improve the error metric without changing the algorithm to improve the situation in various ways.

1. you can calculate 'normals' per vertex by averaging the normals of the two lines connected to it. once you start removing vectors, you do a dot between the normals of the removed vertices and the normal of the line that substitutes those. a vector line that follows a pixel line will have a similar normal as long as you remove vertices within two vital corners. but if you remove a vital corner, you'll change the line normal and even if you do that slightly, accumulated across the whole lines the error will be big.

2. you can weight vertex error by the distance to the vertices of your original shapes (that you've used to draw those.

3. you can try to detect 'vital' edges by calculating the 'hardness' between two neighboring vertex normals. basically, if the angle is beyond a specific treshold, you define that edge as 'vital' and never remove it.

4. more complex error metric would be to draw the reduced shape again. removing some vertices that aren't important will result in a very similar shape, removing 'vital' points will lead to a different bitmap (of course, that's time consuming, but might be ok for an offline process).

alternative: while drawing the original shape, you could first draw the vertices as bigger dots. (different color). then the shapes. all original and important vertices will be surrounded by the dot-colors. so it would be easy to detect those. all vertices within the shape will be overdrawn.

### #12CGameProgrammer  Members

Posted 08 January 2014 - 01:00 PM

Actually I was thinking more about your original idea and I think it would be the solution after all. I was thinking that vertices would be eliminated if the length error of their absence is within a certain margin, but the actual solution (and what I believe you were trying to say) is to only remove the single least useful vertex after evaluating all of them, and then repeat, until the error from removing any remaining vertex is above the threshold. As the least useful vertices are removed, the error from removing the remaining ones becomes much larger and so they would not be erroneously removed.

Last night I wrote the code to complete step #2 (creating the ordered lists of points for each polygon) so today I will write the simplification algorithm and let you know how it goes.

~CGameProgrammer( );
Developer Image Exchange -- New Features: Upload screenshots of your games (size is unlimited) and upload the game itself (up to 10MB). Free. No registration needed.

### #13CGameProgrammer  Members

Posted 13 January 2014 - 11:50 AM

Update: I had implemented the length-based solution but found that it did not work perfectly; it seemed to eliminate curve definition too much or it allowed unnecessary pixels to remain. I came up with a different technique: for each pixel, I find the longest chain of pixels following it that are near the line drawn from the first pixel to the last of the chain. Then I remove all the pixels between the start and end. This nicely handles all sloped lines and can precisely determine the corner pixels.

I'll post code and sample images later. The code is really sloppy right now but I'll try to clean it up a bit first.

~CGameProgrammer( );
Developer Image Exchange -- New Features: Upload screenshots of your games (size is unlimited) and upload the game itself (up to 10MB). Free. No registration needed.

### #14/ Nana111   Members

Posted 13 January 2014 - 09:38 PM

HI there
I have encountered the similar questions with you.I tried to set the raster
image of polygon using this method:

public static PolygonAnnotation CreatePolygonAnnotation(LinePoint[] points);
public static PolygonAnnotation CreatePolygonAnnotation(LinePoint[] points, AnnotationPen outline, AnnotationBrush fill);
public static PolygonAnnotation CreatePolygonAnnotation(LinePoint[] points, AnnotationPen outline, AnnotationBrush fill,
public static PolygonAnnotation CreatePolygonAnnotation(LinePoint[] points, AnnotationPen outline, AnnotationBrush fill,
But it can not work.I want to know that if there is a powerful image toolkit which supports to process raster image of polygon directly.Thanks for any suggestions.

### #15CGameProgrammer  Members

Posted 14 January 2014 - 03:01 PM

I'm too lazy right now to make the code neatly arranged so attached is the quick'n'dirty code. It's a C# console application that reads an image, figures out the required points, and outputs an image showing them. It also outputs intermediate images showing the steps taken to get to that point. I've also attached one of these sample output images.

EDIT: Apparently it doesn't let me attach source code files. Well here is the entire file:

using System;
using System.Collections.Generic;
using System.Drawing;
using System.Drawing.Imaging;

namespace Vector
{
struct Point
{
public static readonly Point Zero = new Point(0,0);

public double X;
public double Y;

public Point (double x, double y)
{
X = x;
Y = y;
}

public override string ToString ()
{
return "{" + X + "," + Y + "}";
}
}

class Polygon
{
public List<Point> Points;

public Polygon ()
{
Points = new List<Point>();
}
}

class Path
{
public List<Point> Points;
public bool[,] EdgeMap;

public Path (bool[,] edgeMap, IEnumerable<Point> points)
{
EdgeMap = (bool[,])edgeMap.Clone();

Points = new List<Point>();
}
}

class Program
{
static readonly double EDGE_OFFSET_THRESHOLD = 1;
static readonly string INPUT_PATH = @"A:\Data\shape.png";

static void Main (string[] args)
{
Bitmap bmp = (Bitmap) Image.FromFile(INPUT_PATH);
bool[,] edgeMap;
int edgeCount;

FindEdges(bmp, out edgeMap, out edgeCount);

Bitmap output = new Bitmap(bmp.Width, bmp.Height);

for (int y = 0; y < bmp.Height; y++)
for (int x = 0; x < bmp.Width; x++)
{
if (edgeMap[x,y])
output.SetPixel(x, y, Color.White);
else
output.SetPixel(x, y, Color.Black);
}

output.Save(INPUT_PATH + ".edges.png", ImageFormat.Png);

// Now find the polygons:

List<Polygon> polygons = new List<Polygon>();

while (edgeCount > 0)
{
// Remove any point to start a new polygon
Point point = FindRandomEdge(edgeMap);
edgeMap[(int) point.X, (int) point.Y] = false;

// Create the first path
Path path = new Path(edgeMap, new Point[0]);
List<Path> paths = new List<Path>();

EvaluatePath(path, paths);

// Now look through the list of paths to find the one(s) with the
// most points. There may be two with the same number of points;
// just choose the first one arbitrarily in that case.

Path bestPath = null;
int bestCount = 0;

foreach (Path option in paths)
{
Point first = option.Points[0];
Point last = option.Points[option.Points.Count-1];

// Look for the path that has all the points and that
// correctly forms a closed polygon.

if (first.X >= last.X-1 &&
first.X <= last.X+1 &&
first.Y >= last.Y-1 &&
first.Y <= last.Y+1 &&
option.Points.Count > bestCount)
{
bestCount = option.Points.Count;
bestPath = option;
}
}

edgeCount -= bestCount;

// Create the new polygon
Polygon polygon = new Polygon();

edgeMap = bestPath.EdgeMap;
}

output = new Bitmap(bmp.Width, bmp.Height);

for (int y = 0; y < bmp.Height; y++)
for (int x = 0; x < bmp.Width; x++)
output.SetPixel(x, y, Color.Black);

for (int n = 0; n < polygons.Count; n++)
{
for (int p = 0; p < polygons[n].Points.Count; p++)
output.SetPixel((int) polygons[n].Points[p].X, (int) polygons[n].Points[p].Y, GetPolygonColor(n));
}

output.Save(INPUT_PATH + ".polygons.png", ImageFormat.Png);

// Now we need to remove useless edge pixels until we only have the corners remaining.

foreach (Polygon polygon in polygons)
{
int p = 0;
while (p < polygon.Points.Count)
{
Point start = polygon.Points[p];
int removeCount = 0;

for (int n = 2; n < polygon.Points.Count; n++)
{
int endIndex = (p+n) % polygon.Points.Count;
Point end = polygon.Points[endIndex];
bool removeToEnd = true;

// Now see if all pixels between start and end are within the allowed distance
// from the line from start to end. If so then keep going. Otherwise we are done
// and can eliminate any pixels between start and end.

for (int m = 1; m < n; m++)
{
int middleIndex = (p+m) % polygon.Points.Count;
Point middle = polygon.Points[middleIndex];
double distance = GetPointSegmentDistance(middle, start, end);
if (distance > EDGE_OFFSET_THRESHOLD)
{
// We've found a pixel that must not be removed so stop here.
removeToEnd = false;
break;
}
}

if (removeToEnd)
removeCount = n-1;
else
break;
}

if (removeCount > 0)
{
int firstRemoval = (p+1) % polygon.Points.Count;
int lastRemoval = (p+removeCount) % polygon.Points.Count;

// We've found at least one pixel to remove.
if (firstRemoval <= lastRemoval)
{
polygon.Points.RemoveRange(firstRemoval, 1+lastRemoval-firstRemoval);
p++;
}
else
{
polygon.Points.RemoveRange(firstRemoval, polygon.Points.Count - firstRemoval);
polygon.Points.RemoveRange(0, 1+lastRemoval);
break;
}
}
else
p++;
}
}

for (int n = 0; n < polygons.Count; n++)
{
for (int p = 0; p < polygons[n].Points.Count; p++)
output.SetPixel((int) polygons[n].Points[p].X, (int) polygons[n].Points[p].Y, Color.White);
}

output.Save(INPUT_PATH + ".corners.png", ImageFormat.Png);
}

static void FindEdges (Bitmap bmp, out bool[,] edgeMap, out int edgeCount)
{
edgeMap = new bool[bmp.Width, bmp.Height];
edgeCount = 0;

// Important: do not count diagonals.

for (int y = 1; y < bmp.Height-1; y++)
{
for (int x = 1; x < bmp.Width-1; x++)
{
if (bmp.GetPixel(x, y).R > 0)
{
// This is a foreground pixel (part of the polygon).
// Count how many surrounded pixels are the background.

int count = (bmp.GetPixel(x-1, y).R == 0 ? 1 : 0) +
(bmp.GetPixel(x+1, y).R == 0 ? 1 : 0) +
(bmp.GetPixel(x, y-1).R == 0 ? 1 : 0) +
(bmp.GetPixel(x, y+1).R == 0 ? 1 : 0);

if (count > 0 && count < 4)
{
// It counts as an edge
edgeMap[x, y] = true;
edgeCount++;
}
}
}
}

// Now we have to eliminate any edge pixels that are adjacent (including diagonally)
// to only one other edge pixel; they are dead-ends.

bool outliersFound = true;
while (outliersFound)
{
outliersFound = false;

for (int y = 1; y < bmp.Height-1; y++)
{
for (int x = 1; x < bmp.Width-1; x++)
{
if (edgeMap[x, y])
{
int count = (edgeMap[x-1, y-1] ? 1 : 0) +
(edgeMap[x-1, y] ? 1 : 0) +
(edgeMap[x-1, y+1] ? 1 : 0) +
(edgeMap[x, y-1] ? 1 : 0) +
(edgeMap[x, y+1] ? 1 : 0) +
(edgeMap[x+1, y-1] ? 1 : 0) +
(edgeMap[x+1, y] ? 1 : 0) +
(edgeMap[x+1, y+1] ? 1 : 0);

if (count == 1)
{
edgeMap[x, y] = false;
edgeCount--;
outliersFound = true;
}
}
}
}
}
}

static double GetPointLineDistance (Point p, Point lineA, Point lineB)
{
double length = GetLength(lineA, lineB);
double xAP = lineA.X - p.X;
double yAP = lineA.Y - p.Y;
double xBA = (lineB.X - lineA.X) / length;
double yBA = (lineB.Y - lineA.Y) / length;

return (xAP * yBA) - (yAP * xBA);
}

static double GetNearestPointOffsetOnLine (Point p, Point lineA, Point lineB)
{
double xAP = p.X - lineA.X;
double yAP = p.Y - lineA.Y;
double xAB = lineB.X - lineA.X;
double yAB = lineB.Y - lineA.Y;

double ABdotAB = (xAB*xAB) + (yAB*yAB);
double APdotAB = (xAP*xAB) + (yAP*yAB);

if (ABdotAB != 0)
return APdotAB / ABdotAB;
else
return double.NaN;
}

static Point GetPointOnLine (Point lineA, Point lineB, double offset)
{
double x = lineA.X + (offset * (lineB.X - lineA.X));
double y = lineA.Y + (offset * (lineB.Y - lineA.Y));
return new Point(x, y);
}

static double GetPointSegmentDistance (Point p, Point segmentA, Point segmentB)
{
double offset = GetNearestPointOffsetOnLine(p, segmentA, segmentB);
if (offset <= 0)
return GetLength(p, segmentA);
else if (offset >= 1)
return GetLength(p, segmentB);
else
return GetLength(p, GetPointOnLine(segmentA, segmentB, offset));
}

static double GetLength (Point p1, Point p2)
{
double dx = p1.X - p2.X;
double dy = p1.Y - p2.Y;

return Math.Sqrt((dx*dx) + (dy*dy));
}

static Color GetPolygonColor (int index)
{
switch (index)
{
case 0:
return Color.FromArgb(64, 0, 0);
case 1:
return Color.FromArgb(64, 64, 0);
case 2:
return Color.FromArgb(0, 64, 0);
case 3:
return Color.FromArgb(0, 64, 64);
case 4:
return Color.FromArgb(0, 0, 64);
case 5:
return Color.FromArgb(64, 0, 64);
default:
return Color.FromArgb(64, 64, 64);
}
}

static void EvaluatePath (Path path, IList<Path> allPaths)
{
while (true)
{
{
// Multiple options so create new paths for each option.
for (int a = 0; a < adjacent.Count; a++)
{
Path branch = new Path(path.EdgeMap, path.Points);

EvaluatePath(branch, allPaths);
}

// This path must be abandoned now.
break;
}
{
// Only one direction to go.
}
else
{
// No more adjacent pixels; we're done with this path
break;
}
}
}

static Point FindRandomEdge (bool[,] edgeMap)
{
for (int y = 1; y < edgeMap.GetLength(1)-1; y++)
for (int x = 1; x < edgeMap.GetLength(0)-1; x++)
{
if (edgeMap[x, y])
return new Point(x, y);
}

return Point.Zero;
}

static List<Point> FindAdjacentEdges (bool[,] edgeMap, Point point)
{
List<Point> list = new List<Point>();

for (int y = (int) point.Y-1; y <= (int) point.Y+1; y++)
for (int x = (int) point.X-1; x <= (int) point.X+1; x++)
{
if (x == point.X && y == point.Y)
continue;

if (edgeMap[x, y])
}

return list;
}
}
}



#### Attached Thumbnails

~CGameProgrammer( );
Developer Image Exchange -- New Features: Upload screenshots of your games (size is unlimited) and upload the game itself (up to 10MB). Free. No registration needed.

Old topic!

Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.