iam dealing with multiscale images , 3 sizes : original , half & double i extract the connected components from each of them separately then do some computations
finally i want to map the bounding rectangles of the half & double scale connected components to their original location and size in the original image
here's my code of resizing and relocating the bounding rectangle
Matrix<int> src = new Matrix<int>(3, 2,3);
Matrix<int> dst = new Matrix<int>(3, 2,2);
IntPtr mat = CvInvoke.cvCreateMat(2, 3, MAT_DEPTH.CV_32F);
src[0,0]=componentBRect.X;
src[0,1]=componentBRect.Y ;
src[1,0]=componentBRect.Right;
src[1,1]=componentBRect.Y;
src[2,0]=componentBRect.X;
src[2,1]=componentBRect.Bottom;
CvInvoke.cvGetAffineTransform(img2, img, mat); //img is the original image & img2 has been resized to 1/2 imgWidth and 1/2 imgHeight
CvInvoke.cvTransform(src.Ptr, dst.Ptr, mat, IntPtr.Zero);
it's supposed to get dst points reszied and relocated but unfortunately the same points are returned again , i think there's a problem with getAffineTransform as it returns an identity matrix
mat=
|1 0 0 |
|0 1 0 | !!
Any Suggestions ?? or is there another way to do what i want without using Affine transformation
Thanks in advance
I have solved it,
I simply found another function - it's cvConvertScale
for (int i = 0; i < Brectangles.Count; i++) {
src[0, 0] = Brectangles[i].X; //top left
src[0, 1] = Brectangles[i].Y;
src[1, 0] = Brectangles[i].Right; //top right
src[1, 1] = Brectangles[i].Y;
src[2, 0] = Brectangles[i].X; //bottom left
src[2, 1] = Brectangles[i].Bottom;
CvInvoke.cvConvertScale(src, dst, rescaleFactor, 0); // rescaleFactor variable is 2 for half size & 0.5 for double size
temp.X = dst[0, 0];
temp.Y = dst[0, 1];
temp.Width = dst[1, 0] - dst[0, 0];
temp.Height = dst[2, 1] - dst[0, 1];
Brectangles[i] = temp;
}
It resizes the rectangle and relocates it in its proper position is the new size