1、递归填充
考虑四连通区域,坐标系x轴朝右为正,y轴朝下为正
void
Fill4(CPoint point,
int
nOldColor,
int
nNewColor)
... {
if(GetPixel(point) == nOldColor)
...{
//填充左边的点
CPoint ptLeft = point;
ptLeft.x = point.x - 1;
if(GetPixel(ptLeft) == nOldColor)
...{
Fill4(ptLeft, nOldColor, nNewColor);
}
//填充右边的点
CPoint ptRight = point;
ptRight.x = point.x + 1;
if(GetPixel(ptRight) == nOldColor)
...{
Fill4(ptRight, nOldColor, nNewColor);
}
//填充下边的点
CPoint ptBottom = point;
ptBottom.y = point.y + 1;
if(GetPixel(ptBottom) == nOldColor)
...{
Fill4(ptBottom, nOldColor, nNewColor);
}
//填充右边的点
CPoint ptTop = point;
ptTop.y = point.y - 1;
if(GetPixel(ptTop) == nOldColor)
...{
Fill4(ptTop, nOldColor, nNewColor);
}
}
}
... {
if(GetPixel(point) == nOldColor)
...{
//填充左边的点
CPoint ptLeft = point;
ptLeft.x = point.x - 1;
if(GetPixel(ptLeft) == nOldColor)
...{
Fill4(ptLeft, nOldColor, nNewColor);
}
//填充右边的点
CPoint ptRight = point;
ptRight.x = point.x + 1;
if(GetPixel(ptRight) == nOldColor)
...{
Fill4(ptRight, nOldColor, nNewColor);
}
//填充下边的点
CPoint ptBottom = point;
ptBottom.y = point.y + 1;
if(GetPixel(ptBottom) == nOldColor)
...{
Fill4(ptBottom, nOldColor, nNewColor);
}
//填充右边的点
CPoint ptTop = point;
ptTop.y = point.y - 1;
if(GetPixel(ptTop) == nOldColor)
...{
Fill4(ptTop, nOldColor, nNewColor);
}
}
}
递归算法效率比较低,当点的数量比较大的时候就能看出来速度慢了,在VC环境下默认的栈大小是1M,在project->setting->link的output中可以修改栈的大小,/statck:size
2、非递归算法
原理是一样的,种子象素先入栈,然后找上下左右,如此循环,但效率就高多了,本来我用的是CArray来保存这些点,但点的数量有百万的时候CArray的处理速度也跟不上了,这里我开辟了一块足够大的空间,然后在这块内存里操作,速度就快多了,但是会造成空间浪费,但用完之后就删掉了
void
Fill4(CPoint point,
int
nOldColor,
int
nNewColor)
... {
CPoint* pPoint = new CPoint[ENOUGHSIZE];
memset(pPoint, 0, sizeof(CPoint) * ENOUGHSIZE);
//种子象素入栈
int nNum = 0;
pPoint[nNum++] = point;
int i = 0;
while(i <= nNum)
...{
//部分省略
//填充左边的点
if(GetPixel(ptLeft) == nOldColor)
...{
SetPixel(ptLeft, nNewColor);
pPoint[nNum++] = ptLeft;
}
//填充右边的点
......
//填充上边的点
......
//填充下边的点
......
i++;
}
if(pPoint != NULL)
...{
delete[] pPoint;
pPoint = NULL;
}
}
... {
CPoint* pPoint = new CPoint[ENOUGHSIZE];
memset(pPoint, 0, sizeof(CPoint) * ENOUGHSIZE);
//种子象素入栈
int nNum = 0;
pPoint[nNum++] = point;
int i = 0;
while(i <= nNum)
...{
//部分省略
//填充左边的点
if(GetPixel(ptLeft) == nOldColor)
...{
SetPixel(ptLeft, nNewColor);
pPoint[nNum++] = ptLeft;
}
//填充右边的点
......
//填充上边的点
......
//填充下边的点
......
i++;
}
if(pPoint != NULL)
...{
delete[] pPoint;
pPoint = NULL;
}
}
扫描线填充算法的效率比较高,但是对于任意图形,还不知道怎么个写法,望知道的朋友赐教,先谢谢了