opencv中图像失焦检测

失焦的图片和对焦准确的图片最大的区别就是正常图片轮廓明显，而失焦图片几乎没有较大像素值之间的变化

对图像的横向，以及纵向，分别做差分，累计差分可以用来作为判断是否失焦的参考

两个函数，一个简单粗暴直接根据差分值判断是否失焦，适合确定样本类型的情况，另外一种，需要进一步判断

//简单设定阈值判断是否失焦
bool focusDetect(Mat& img){

clock_t start, end;
start = clock();
int diff = 0;
int diff_thre = 20;
int diff_sum_thre = 1000;
for (int i = img.rows / 10; i < img.rows; i += img.rows / 10){
uchar* ptrow = img.ptr<uchar>(i);
for (int j = 0; j < img.cols - 1; j++){
if (abs(ptrow[j + 1] - ptrow[j])>diff_thre)
diff += abs(ptrow[j + 1] - ptrow[j]);
}
cout << diff << endl;
}
end = clock();
cout << "time=" << end - start << endl;

bool res = true;
if (diff < diff_sum_thre) {
cout << "the focus might be wrong!" << endl;
res = false;
}

return res;
}

//返回一个与焦距是否对焦成功的一个比例因子
double focus_measure_GRAT(Mat Image)
{
double threshold = 0;
double temp = 0;
double totalsum = 0;
int totalnum = 0;

for (int i=0; i<Image.rows; i++)
{
uchar* Image_ptr = Image.ptr<uchar>(i);
uchar* Image_ptr_1 = Image.ptr<uchar>(i+1);
for (int j=0; j<Image.cols; j++)
{
temp = max(abs(Image_ptr_1[j]-Image_ptr[j]), abs(Image_ptr[j+1]-Image_ptr[j]));
totalsum += temp;
totalnum += 1;
}
}

double FM = totalsum/totalnum;

return FM;
}