Image Thinning

Hi Folks,
I am still working on my degree project of fingerprint matching. I have changed the image of the print so far in the following ways; grey-scale, smooth, Binarized it and now I am thinning it i.e making it one pixel in width (skeleton). I will then be extracting the minutiae points and comparing them to the prints on file. However, I am having some trouble trying to get the results I need with thinning. I have some code to thin an image but instead of thinning it it appears to be making it larger?? I have stepped through it and believe it should work for me. If anyone has experience of thinning I would be grateful for any suggestions.
Thanks in advance,
-Colin.

Code:

---

void __fastcall TForm1::ThinningClick(TObject *Sender)
{
  //THINNING
  int count,count2,row,col;
  int num_transitions ;
  int transitions ;
  int tempimage[297][353]={0};
  int row1,col1;
  for(row = 1; row < Image1->Height-1 ; row++)
    {
    for(col = 1; col < Image1->Width-1 ; col++)
      {
      count = 0;
      num_transitions = 0;
      // check  for N(p)
      if (Image1->Canvas->Pixels[col][row] == 0xFFFFFF)
        {
        if (Image1->Canvas->Pixels[col-1][row-1] != 0)
          count++;
        if (Image1->Canvas->Pixels[col][row-1] != 0)
          count++;
        if (Image1->Canvas->Pixels[col+1][row-1] != 0)
          count++;
        if (Image1->Canvas->Pixels[col+1][row] != 0)
          count++;
        if (Image1->Canvas->Pixels[col-1][row] != 0)
          count++;
        if (Image1->Canvas->Pixels[col+1][row+1] != 0)
          count++;
        if (Image1->Canvas->Pixels[col][row+1] != 0)
          count++;
        if (Image1->Canvas->Pixels[col-1][row+1] != 0)
          count++;
        if (count != 8)
          {
          // 2 <= N(p) <= 6
          if (count >= 2 && count <= 6)
            {
            if(Image1->Canvas->Pixels[col-1][row] == 0 &&
              Image1->Canvas->Pixels[col-1][row+1] == 0xFFFFFF)
              num_transitions++ ;
            if(Image1->Canvas->Pixels[col-1][row+1] == 0 &&
              Image1->Canvas->Pixels[col][row+1] == 0xFFFFFF)
              num_transitions++ ;
            if(Image1->Canvas->Pixels[col][row+1] == 0 &&
              Image1->Canvas->Pixels[col+1][row+1] == 0xFFFFFF)
              num_transitions++ ;
            if(Image1->Canvas->Pixels[col+1][row+1] == 0 &&
              Image1->Canvas->Pixels[col+1][row] == 0xFFFFFF)
              num_transitions++ ;
            if(Image1->Canvas->Pixels[col+1][row] == 0 &&
              Image1->Canvas->Pixels[col+1][row-1] == 0xFFFFFF)
              num_transitions++ ;
            if(Image1->Canvas->Pixels[col+1][row-1] == 0 &&
              Image1->Canvas->Pixels[col][row-1] == 0xFFFFFF)
              num_transitions++ ;
            if(Image1->Canvas->Pixels[col][row-1] == 0 &&
              Image1->Canvas->Pixels[col-1][row-1] == 0xFFFFFF)
              num_transitions++ ;
            if(Image1->Canvas->Pixels[col-1][row-1] == 0 &&
              Image1->Canvas->Pixels[col-1][row] == 0xFFFFFF)
              num_transitions++ ;
            //S(p) = 1
            if (num_transitions == 1)
              {
              // if p2 * p4 * p6 = 0
              if (Image1->Canvas->Pixels[col][row-1] == 0 || 
                Image1->Canvas->Pixels[col][row+1] == 0 ||
                Image1->Canvas->Pixels[col+1][row] == 0)
                {
                // if p4 * p6 * p8 = 0
                if(Image1->Canvas->Pixels[col][row+1] == 0 ||
                  Image1->Canvas->Pixels[col+1][row] == 0 ||
                  Image1->Canvas->Pixels[col][row-1] == 0)
                  tempimage[col][row] = 0;
                else
                  tempimage[col][row] =  0xFFFFFF;
                  }
              else
                tempimage[col][row] = 0xFFFFFF;
              }
            else
              tempimage[col][row] =  0xFFFFFF;
            }
          else
            tempimage[col][row] =  0xFFFFFF;
          }
        else
          tempimage[col][row] =  0xFFFFFF;
        }
      else
        tempimage[col][row] =0;//0xFFFFFF;
      }
    }
  //copy thinned image back to original

    for(row = 0; row < Image1->Height ; row++)
      for(col = 0; col < Image1->Width; col++)
      Image1->Canvas->Pixels[col][row]=tempimage[col][row];

  // step 2 of the thinning algorithm

  for(row = 1; row < Image1->Height-1 ; row++)
    {
    for(col = 1; col < Image1->Width-1 ; col++)
      {
      count2 = 0;
      transitions = 0;
      if (Image1->Canvas->Pixels[col][row] == 0xFFFFFF)
        {
        if (Image1->Canvas->Pixels[col-1][row-1] != 0)
          count2++;
        if (Image1->Canvas->Pixels[col][row-1] != 0)
          count2++;
        if (Image1->Canvas->Pixels[col+1][row-1] != 0)
          count2++;
        if (Image1->Canvas->Pixels[col+1][row] != 0)
          count2++;
        if (Image1->Canvas->Pixels[col-1][row] != 0)
          count2++;
        if (Image1->Canvas->Pixels[col+1][row+1] != 0)
          count2++;
        if (Image1->Canvas->Pixels[col][row+1] != 0)
          count2++;
        if (Image1->Canvas->Pixels[col-1][row+1] != 0)
          count2++;

        if (count2 != 8)
          {
          if (count2 >= 2 && count2 <= 6)
            {
            if(Image1->Canvas->Pixels[col-1][row] == 0 &&
              Image1->Canvas->Pixels[col-1][row+1] == 0xFFFFFF)
              transitions++ ;
            if(Image1->Canvas->Pixels[col-1][row+1] == 0 &&
              Image1->Canvas->Pixels[col][row+1] == 0xFFFFFF)
              transitions++ ;
            if(Image1->Canvas->Pixels[col][row+1] == 0 &&
              Image1->Canvas->Pixels[col+1][row+1] == 0xFFFFFF)
              transitions++ ;
            if(Image1->Canvas->Pixels[col+1][row+1] == 0 &&
              Image1->Canvas->Pixels[col+1][row] == 0xFFFFFF)
              transitions++ ;
            if(Image1->Canvas->Pixels[col+1][row] == 0 &&
              Image1->Canvas->Pixels[col+1][row-1] == 0xFFFFFF)
              transitions++ ;
            if(Image1->Canvas->Pixels[col+1][row-1] == 0 &&
              Image1->Canvas->Pixels[col][row-1] == 0xFFFFFF)
              transitions++ ;
            if(Image1->Canvas->Pixels[col][row-1] == 0 &&
              Image1->Canvas->Pixels[col-1][row-1] == 0xFFFFFF)
              transitions++ ;
            if(Image1->Canvas->Pixels[col-1][row-1] == 0 &&
              Image1->Canvas->Pixels[col-1][row] == 0xFFFFFF)
              transitions++ ;

            if (transitions == 1)
              {
              // if p2 * p4 * p8 = 0
              if(Image1->Canvas->Pixels[col][row-1] == 0 ||
                Image1->Canvas->Pixels[col][row+1] == 0 ||
                Image1->Canvas->Pixels[col][row-1] == 0)
                {
                // if p2 * p4 * p6
                if(Image1->Canvas->Pixels[col-1][row] == 0 ||
                  Image1->Canvas->Pixels[col+1][row] == 0 ||
                  Image1->Canvas->Pixels[col][row-1] == 0)
                  tempimage[col][row] = 0;
                else
                  tempimage[col][row] =  0xFFFFFF;
                }
              else
                tempimage[col][row] =  0xFFFFFF;
              }
            else
              tempimage[col][row] = 0xFFFFFF;
            }
          else
            tempimage[col][row] =  0xFFFFFF;
          }
        else
          tempimage[col][row] = 0xFFFFFF;
        }
      else
        tempimage[col][row] = 0;
      }
 }

  for(row = 0; row < Image1->Height ; row++)
    for(col = 0; col < Image1->Width; col++)
      Image1->Canvas->Pixels[col][row]=tempimage[col][row];


}

---

It's difficult to look through all that code and find the problem, but I think I understand what you're trying to do, and here's an alternative approach:

Code:

---

loop y from 0 to image y {
x is 0
while x < image x {
  if(image[x] == 0xfffffff) {
      runsize is 0
      while(image[x] == 0xffffffff) {
        ++runsize; ++x;
      }
      clear runsize pixels, turn on single pixel in center of run
  }
}

---

Good luck with your project

Thanks ggs for your reply.
I have been working on your suggestion but am having trouble still. Can you explain "runsize" in your code to me in more detail.
Thanks again for you time,
-Colin

runsize just means the number of pixels in a row that are turned on. you'd use this to find out where the middle pixel is, how many pixels to turn off...