洪水填充算法分析

Question

I've got some method with flood fill algorithm. It is very simple

1. Go to first obstacle on top.

2. change pixels color to the bottom

3. while changing check if left/right pixel is in different color

4. if yes: color this column too (stack.push())

5. loop.

     Stack<Point> st = new Stack<Point>(); bool spLeft, spRight; Bitmap b = canvas.buffer; st.Push(start); spLeft = spRight = false; Point p = new Point(); while (st.Count > 0) { //going as far top as possible (finding first obstacle) p = st.Pop(); while (pY >= 0 && b.GetPixel(pX, pY) == oldColor) pY--; p.Y++; spLeft = spRight = false; //looping on every oldColored pixel in column while (pY < b.Height && b.GetPixel(pX, pY) == oldColor) { b.SetPixel(pX, pY, state.currentColor); //setting new color //checking if left pixel is oldColored and if it doesn't belong to span if (!spLeft && pX > 0 && b.GetPixel(pX - 1, pY) == oldColor) { st.Push(new Point(pX - 1, pY)); spLeft = true; } //checking if left pixel isn't oldColored and if it belongs to span else if (spLeft && pX > 0 && b.GetPixel(pX - 1, pY) != oldColor) { spLeft = false; } if (!spRight && pX < b.Width - 1 && b.GetPixel(pX + 1, pY) == oldColor) { st.Push(new Point(pX + 1, pY)); spRight = true; } else if (spRight && pX < b.Width - 1 && b.GetPixel(pX + 1, pY) != oldColor) { spRight = false; } p.Y++; } } 

The point is that i just do not understand these parts

    //checking if left pixel isn't oldColored and if it belongs to span
    else if (spLeft && p.X > 0 && b.GetPixel(p.X - 1, p.Y) != oldColor) {
    spLeft = false;

and

    else if (spRight && p.X < b.Width - 1 && b.GetPixel(p.X + 1, p.Y) != oldColor) {
    spRight = false;
            }

Without these, code works fine, and it seems like it have the same amount of iterrations. Can you help me to figure if these line are realy useless or I just dont understand them ? (I cant belive my friend put them without purpose)

Answer 1

They allow multiple regions to be filled. The opening if statements checks they are false and adds a pixel to the stack. Those reset when that region has finished.

Without resetting spLeft region 2 would not be filled since it would have been set to true (which avoids add lots to the stack unecesserily) when the first region was encountered.