double sharpnessDetect(const cv::Mat &srcImg) { int gaussianSize = 3; cv::Mat img; if (srcImg.channels() != 1) cv::cvtColor(srcImg, img, COLOR_BGR2GRAY); else img = srcImg; cv::Mat out; cv::GaussianBlur(img, out, cv::Size(gaussianSize, gaussianSize), 0, 0); unsigned long d_Fver = 0, d_Fhor = 0, d_Bver = 0, d_Bhor = 0; unsigned long vver = 0, vhor = 0; unsigned long s_Fver = 0, s_Fhor = 0, s_Vver = 0, s_Vhor = 0; double b_Fver = 0.0, b_Fhor = 0.0; double blur_F = 0.0; for (int r = 0; r < img.rows; r++) { for (int c = 0; c < img.cols; c++) { if (r != 0) d_Fver = abs(img.at<uchar>(r, c) - img.at<uchar>(r - 1, c)) if (c != 0) d_Fhor = abs(img.at<uchar>(r, c) - img.at<uchar>(r, c - 1)) if (r != 0) d_Bver = abs(out.at<uchar>(r, c) - out.at<uchar>(r - 1, c)) if (c != 0) d_Bhor = abs(out.at<uchar>(r, c) - out.at<uchar>(r, c - 1)) vver = (d_Fver - d_Bver > 0) ? (d_Fver - d_Bver) : 0 vhor = (d_Fhor - d_Bhor > 0) ? (d_Fhor - d_Bhor) : 0 s_Fver += d_Fver; s_Fhor += d_Fhor; s_Vver += vver; s_Vhor += vhor; } } b_Fver = (s_Fver - s_Vver) / ((double)s_Fver + 1); b_Fhor = (s_Fhor - s_Vhor) / ((double)s_Fhor + 1); blur_F = (b_Fver > b_Fhor) ? b_Fver : b_Fhor; return 1 - blur_F; }

这段代码是一个双边缘检测函数，用于计算图像的清晰度。该函数首先将输入图像转换为灰度图像（如果输入图像不是单通道的），然后使用高斯模糊对灰度图像进行模糊处理。接下来，该函数计算前景和背景的差异，并使用该差异计算垂直和水平方向的前景梯度和背景梯度。然后，该函数计算每个像素的前景梯度和背景梯度之差，并将其用于计算垂直和水平方向的前景梯度和背景梯度的总和。最后，该函数使用前景梯度和背景梯度的总和计算清晰度值，并将其返回。

需要注意的是，在代码中有一些缺少分号的错误，需要在相应的地方添加分号。另外，该函数的实现可能存在一些性能问题，特别是在处理大型图像时。可以考虑使用多线程或GPU加速等技术来提高性能。