Calculating the bounding rectangle on a binary image - c#

I'm attempting to get the bounds of the edges of interest in a binary image using the following method. Sadly my maths seems to be letting me down and I'm only getting a rectangle 2px smaller in each dimension than the original image.
Can someone show me where I have gone wrong?
Note. FastBitmap is a class that allows fast access to pixel data.
private Rectangle FindBox(Bitmap bitmap, byte indexToRemove)
{
int width = bitmap.Width;
int height = bitmap.Height;
int minX = width;
int minY = height;
int maxX = 0;
int maxY = 0;
using (FastBitmap fastBitmap = new FastBitmap(bitmap))
{
for (int y = 0; y < height; y++)
{
for (int x = 0; x < width; x++)
{
if (fastBitmap.GetPixel(x, y).B == indexToRemove)
{
if (x < minX)
{
minX = x;
}
if (x > maxX)
{
maxX = x;
}
if (y < minY)
{
minY = y;
}
if (y > maxY)
{
maxY = y;
}
}
}
}
}
// check
if ((minX == width) && (minY == height) && (maxX == 0) && (maxY == 0))
{
minX = minY = 0;
}
return new Rectangle(minX, minY, maxX - minX + 1, maxY - minY + 1);
}
The image I'm testing.

You appear to be checking EVERY pixel to see if a match exists in any x and y. Instead, what you want to do is check the minx, maxx, miny, and maxy separately.
For the minY, you want start at the top and check each row down until you hit a y row that has a matching pixel.
For the maxY, you want start at the bottom and check each row up until you hit a y row that has a matching pixel.
For the minX, you want start at the left and check each column until you hit a x column that has a matching pixel.
For the maxX, you want start at the right and check each column until you hit a x column that has a matching pixel.
Something like this:
minY = getMinY(fastBitmap, indexToRemove);
maxY = getMinY(fastBitmap, indexToRemove);
minX = getMinY(fastBitmap, indexToRemove);
maxX = getMinY(fastBitmap, indexToRemove);
int getMinY(Bitmap bitmap, byte indexToRemove)
{
for (int y = 0; y < height; y++)
{
for (int x = 0; x < width; x++)
{
if (fastBitmap.GetPixel(x, y).B == indexToRemove)
{
return y;
}
}
}
return 0;
}
int getMaxY(Bitmap bitmap, byte indexToRemove)
{
for (int y = height; y > 0; y--)
{
for (int x = 0; x < width; x++)
{
if (fastBitmap.GetPixel(x, y).B == indexToRemove)
{
return y;
}
}
}
return height;
}
etc...
You should be able to write the getMinX and getMaxY yourself.

Related

Equality between images c#

I need to compare two images and see if they are the same.
The images only contain text.
I have an image on the hard drive saved and I need to see if it is the same as the one I am downloading.
I tried with this code to establish a margin of error of 5%, but the main problem I have is that the text of image can be rotated between 0 and 30 degrees or 0 and -30
int width;
int height;
if (bmp1.width < bmp2.width)
{
width = bmp1.width;
}
else
{
width = bmp2.width;
}
if (bmp1.heigth < bmp2.width)
{
heigth = bmp1.heigth;
}
else
{
heigth = bmp2.heigth;
}
int contador = 0;
for (int x = 0; x < width; x++)
{
for (int y = 0; y < height; y++)
{
if (bmp1.GetPixel(x, y) == bmp2.GetPixel(x, y))
{
contador++
}
}
}
int similar_percent = (contador / (height * width) * 100);
I leave two small examples of the images

Cut transparent parts image

The updated image:
thanks you very much man but i wanted to do somthing little different, to cut each rectangle here as a sperate image.Lets try first to find the blue block Bounds. Sounds hard but its actually simple.
look when i have done so far:
private unsafe Bitmap CodeImage(Bitmap bmp)
{
Bitmap bmpRes = new Bitmap(bmp.Width, bmp.Height);
BitmapData bmData = bmp.LockBits(new Rectangle(0, 0, bmp.Width, bmp.Height), System.Drawing.Imaging.ImageLockMode.ReadOnly, bmp.PixelFormat);
IntPtr scan0 = bmData.Scan0;
int stride = bmData.Stride;
int nWidth = bmp.Width;
int nHeight = bmp.Height;
int minX = 10000 ;
int maxX = -10000;
int minY = 10000;
var maxY = -10000;
bool found = false;
for (int y = 0; y < nHeight; y++)
{
byte* p = (byte*)scan0.ToPointer();
p += y * stride;
for (int x = 0; x < nWidth; x++)
{
if (p[3]!=0) //Check if pixel is transpert;
{
found = true;
if (x < minX)
minX = x;
if (y < minY)
minY = y;
if (x > maxX)
maxX = x;
if (y > maxY)
maxY = y;
}
else
{
if (found)
{
Rectangle temp = new Rectangle(minX, minY, maxX - minX, maxY-minY);
return bmp.Clone(temp, bmp.PixelFormat);
}
}
p += 4;
}
}
return null;
}
you actually was write and i should calculate the width like this :
int width = maxX - minX; and it actually works.. but the height is 0....
try this out man its outputing almost correct rectangle with these bounds:
(200,800,400, and 0 on the height).
i just used parts of your code in my algorithm and yea you were right but now there is a little problem with the height i' would very appreciate if you will have a look
I'll write this in pseudo-ish-code since you should be able then to apply it to any type of image or language...
var minX = 10000
var maxX = -10000
var minY = 10000
var maxY = -10000
for (var y = 0 to height)
{
for (var x = 0 to width)
{
if (pixel(x,y).Color != transparent)
{
if (x < minX) minX = x
if (y < minY) minY = y
if (x > maxX) maxX = x
if (y > maxY) maxY = y
}
}
}
var cropRectangle = (minX, minY, maxX, maxY)
You can now use standard functions on the bitmap to get that area, which should be the area bounded by the non-transparent pixels.

Map flipped on the horizontal Axis when displayed

I am having a problem with displaying maps on screen. I am giving the developer the choice between square tiles and hexagonal tiles. The map is saved as a .txt
The square tiled map displays correctly as shown
where as the hexagonal map is upside down
I am completely stumped as to how to go about fixing the issue I have supplied the entire function bellow
void displayMap(){
//Draw Tiles on the Screen
if (useSquareTiles == true && useHexagonTiles == false) {
for (int y = 0; y < mapHeight; y++) {
for (int x = 0; x < mapWidth; x++) {
GameObject.Instantiate (mapArray [y] [x], new Vector3 (x, mapHeight - y, 0),
Quaternion.Euler (-90, 0, 0));
}
}
}
float hexWidth = 0.85f; //The width of the HexTiles
if (useHexagonTiles == true && useSquareTiles == false) {
for (int y = 0; y < mapHeight; y++) {
for (int x = 0; x < mapWidth; x++) {
GameObject.Instantiate (mapArray[y][x], new Vector3 (x * hexWidth, y + (0.5f * Mathf.Abs (x) % 1), 0),
Quaternion.Euler (-90, 0, 0));
}
}
}
}

c# auto-crop an image

I have a black and white System.Drawing.Bitmap I need to auto-crop it so that it is only as big as needed to fit the image. This image always starts at the top left (0,0) position but I'm not sure how much height and width is will require. If there any way to auto-crop it to size?
The following is the code I used:
// Figure out the final size
int maxX = 0;
int maxY = 0;
for (int x = 0; x < bitmap.Width; x++)
{
for (int y = 0; y < bitmap.Height; y++)
{
System.Drawing.Color c = bitmap.GetPixel(x, y);
System.Drawing.Color w = System.Drawing.Color.White;
if (c.R != w.R || c.G != w.G || c.B != w.B)
{
if (x > maxX)
maxX = x;
if (y > maxY)
maxY = y;
}
}
}
maxX += 2;
If your image does not start at 0,0 like mine, here is a variant that gets you the auto-crop bounds:
public static Rectangle GetAutoCropBounds(Bitmap bitmap)
{
int maxX = 0;
int maxY = 0;
int minX = bitmap.Width;
int minY = bitmap.Height;
for (int x = 0; x < bitmap.Width; x++)
{
for (int y = 0; y < bitmap.Height; y++)
{
var c = bitmap.GetPixel(x, y);
var w = Color.White;
if (c.R != w.R || c.G != w.G || c.B != w.B)
{
if (x > maxX)
maxX = x;
if (x < minX)
minX = x;
if (y > maxY)
maxY = y;
if (y < minY)
minY = y;
}
}
}
maxX += 2;
return new Rectangle(minX, minY, maxX - minX, maxY - minY);
}
Note: if you are processing any amount of images in quantity or high-resolution then you'll want to look into faster alternatives than the GetPixel method.

Calculate pixels within a polygon

In an assignment for school do we need to do some image recognizing, where we have to find a path for a robot.
So far have we been able to find all the polygons in the image, but now we need to generate a pixel map, that be used for an astar algorithm later. We have found a way to do this, show below, but the problem is that is very slow, as we go though each pixel and test if it is inside the polygon. So my question is, are there a way that we can generate this pixel map faster?
We have a list of coordinates for the polygon
private List<IntPoint> hull;
The function "getMap" is called to get the pixel map
public Point[] getMap()
{
List<Point> points = new List<Point>();
lock (hull)
{
Rectangle rect = getRectangle();
for (int x = rect.X; x <= rect.X + rect.Width; x++)
{
for (int y = rect.Y; y <= rect.Y + rect.Height; y++)
{
if (inPoly(x, y))
points.Add(new Point(x, y));
}
}
}
return points.ToArray();
}
Get Rectangle is used to limit the search, se we don't have to go thoug the whole image
public Rectangle getRectangle()
{
int x = -1, y = -1, width = -1, height = -1;
foreach (IntPoint item in hull)
{
if (item.X < x || x == -1)
x = item.X;
if (item.Y < y || y == -1)
y = item.Y;
if (item.X > width || width == -1)
width = item.X;
if (item.Y > height || height == -1)
height = item.Y;
}
return new Rectangle(x, y, width-x, height-y);
}
And atlast this is how we check to see if a pixel is inside the polygon
public bool inPoly(int x, int y)
{
int i, j = hull.Count - 1;
bool oddNodes = false;
for (i = 0; i < hull.Count; i++)
{
if (hull[i].Y < y && hull[j].Y >= y
|| hull[j].Y < y && hull[i].Y >= y)
{
try
{
if (hull[i].X + (y - hull[i].X) / (hull[j].X - hull[i].X) * (hull[j].X - hull[i].X) < x)
{
oddNodes = !oddNodes;
}
}
catch (DivideByZeroException e)
{
if (0 < x)
{
oddNodes = !oddNodes;
}
}
}
j = i;
}
return oddNodes;
}
There are some interesting discussion here on polygon hit tests, but it sounds to me as if you might be better off with a polygon fill.
You may want to look for a Plygon Triangulation algorithm.
Also, note that catching an exception is far more time-consuming that checking the right condition. So i suggest you to convert your existing code in:
public bool inPoly(int x, int y)
{
int i, j = hull.Count - 1;
var oddNodes = false;
for (i = 0; i < hull.Count; i++)
{
if (hull[i].Y < y && hull[j].Y >= y
|| hull[j].Y < y && hull[i].Y >= y)
{
var delta = (hull[j].X - hull[i].X);
if (delta == 0)
{
if (0 < x) oddNodes = !oddNodes;
}
else if (hull[i].X + (y - hull[i].X) / delta * delta < x)
{
oddNodes = !oddNodes;
}
}
j = i;
}
return oddNodes;
}

Categories