CImg* CDimageView::Dwtstep(CImg* pImg, int Inv)
{
// 初始化目标图像
CImg* mImg = new CImg(*pImg);
mImg->InitPixels(0);
int width=pImg->GetWidthPixel();
int height=pImg->GetHeight();
int nMaxWLevel = Log2(width);
int nMaxHLevel = Log2(height);
int nMaxLevel;
if (width == 1<<nMaxWLevel && height == 1<<nMaxHLevel)
nMaxLevel = min(nMaxWLevel, nMaxHLevel);
// 如果小波变换的存储内存还没有分配，则分配此内存,用double保存graylevel
if(!Dwttemp){
Dwttemp = new double[width*height];
// 将图象数据放入m_pDbImage中
for (int i=0; i<height; i++)
{
for (int j=0; j<width; j++)
Dwttemp[i * width + j] = double(pImg->GetGray(j,i));
}
}
if (!DWTStep_2D(Dwttemp, nMaxWLevel-Curdpth, nMaxHLevel-Curdpth,nMaxWLevel, nMaxHLevel, Inv))
return FALSE;
if (Inv)
Curdpth --;
// 否则加1
else
Curdpth ++;
BYTE gray;
//将double里gray数据转入24位bmp(gray,gray,gray)
int lfw = width>>Curdpth, lfh = height>>Curdpth; //尺度区域
for (int i=0; i<height; i++)
{
for (int j=0; j<width; j++){
if(i<lfh && j<lfw){ //低频X低频，>0直接转
gray = BYTE(Dwttemp[i * width + j]);
mImg->SetPixel(j,i,RGB(gray,gray,gray));
}
else{ //小于零部分 +128显示
gray = BYTE(Dwttemp[i * width + j]+128);
mImg->SetPixel(j,i,RGB(gray,gray,gray));
}
}
}
return mImg;
}
CImg CImg::operator ! ()
{
CImg grayRet = *this;
grayRet.InitPixels(255); //结果图像置白
int nHeight = GetHeight();
int nWidth = GetWidthPixel();
int i,j;
for(i=0; i<nHeight; i++)
{
for(j=0; j<nWidth; j++)
{
int pixel = 255 - GetGray(j, i);
grayRet.SetPixel(j, i, RGB(pixel, pixel, pixel));
}
}
return grayRet;
}
CImg* CDimageView::Canny(CImg* pImg)
{
// 各方向梯度值
// 使用Prewitt模板计算各个方向上的梯度值
CImg* imgGH=PrewittEdge(pImg,1);
CImg* imgGV=PrewittEdge(pImg,2);
CImg* imgGCW=PrewittEdge(pImg,3);
CImg* imgGCCW=PrewittEdge(pImg,4);
CImg* imgGratitude = new CImg(*pImg);
imgGratitude->InitPixels(0);
int width=pImg->GetWidthPixel();
int height=pImg->GetHeight();
// 最大梯度方向
BYTE * pbDirection = new BYTE [height * width];
memset(pbDirection, 0,height * width * sizeof(BYTE));
// 寻找每点的最大梯度方向并写入对应的最大梯度值
for (int i=0; i<height; i++)
{
for (int j=0; j<width; j++)
{
BYTE gray = 0;
if (imgGH->GetGray(j, i) > gray)
{
gray = imgGH->GetGray(j, i);
pbDirection[i * width + j] = 1;
imgGratitude->SetPixel(j, i, RGB(gray, gray, gray));
}
if (imgGV->GetGray(j, i) > gray)
{
gray = imgGV->GetGray(j, i);
pbDirection[i * width + j] = 2;
imgGratitude->SetPixel(j, i, RGB(gray, gray, gray));
}
if (imgGCW->GetGray(j, i) > gray)
{
gray = imgGCW->GetGray(j, i);
pbDirection[i * width + j] = 3;
imgGratitude->SetPixel(j, i, RGB(gray, gray, gray));
}
if (imgGCCW->GetGray(j, i) > gray)
{
gray = imgGCCW->GetGray(j, i);
pbDirection[i * width + j] = 4;
imgGratitude->SetPixel(j, i, RGB(gray, gray, gray));
}
}
}
// 阈值化时重用前面的对象
CImg *pImgThreL = imgGH, *pImgThreH = imgGV;
// 检查阈值参数，如未给出阈值则计算以取得最佳阈值
int bThreH;
int bThreL;
bThreH = 1.2 * imgGratitude->DetectThreshold(100);
bThreL = 0.4 * bThreH;
// 将最大梯度图像按高低值分别进行阈值化
imgGratitude->Threshold(pImgThreL, bThreL);
imgGratitude->Threshold(pImgThreH, bThreH);
// 初始化目标图像
CImg* mImg = new CImg(*pImg);
mImg->InitPixels(0);
// 根据低阈值图像在高阈值图像上进行边界修补
for (int i=1; i<height-1; i++)
{
for (int j=1; j<width-1; j++)
{
if (pImgThreH->GetGray(j, i))
{
// 高阈值图像上发现点直接确定
mImg->SetPixel(j, i, RGB(255, 255, 255));
// 搜索梯度最大方向上的邻域
switch ( pbDirection[i * width + j] ) {
case 1: // 水平方向
if (pImgThreL->GetGray(j+1, i))
{
pImgThreH->SetPixel(j+1, i, RGB(255, 255, 255));
}
if (pImgThreL->GetGray(j-1, i))
{
pImgThreH->SetPixel(j-1, i, RGB(255, 255, 255));
}
break;
case 2: // 垂直方向
if (pImgThreL->GetGray(j, i+1))
{
pImgThreH->SetPixel(j, i+1, RGB(255, 255, 255));
}
if (pImgThreL->GetGray(j, i-1))
{
pImgThreH->SetPixel(j, i-1, RGB(255, 255, 255));
}
break;
case 3: // 45度方向
if (pImgThreL->GetGray(j+1, i-1))
{
pImgThreH->SetPixel(j+1, i-1, RGB(255, 255, 255));
}
if (pImgThreL->GetGray(j-1, i+1))
{
pImgThreH->SetPixel(j-1, i+1, RGB(255, 255, 255));
}
break;
case 4: // 135度方向
if (pImgThreL->GetGray(j+1, i+1))
{
pImgThreH->SetPixel(j+1, i+1, RGB(255, 255, 255));
}
if (pImgThreL->GetGray(j-1, i-1))
{
pImgThreH->SetPixel(j-1, i-1, RGB(255, 255, 255));
}
break;
}
}//if
}//for j
}//for i
delete pbDirection;
return mImg;
}