int detecCircles(cv::Mat image, vector<cv::Vec3f> &circles, vector<cv::Point> &bias)
{
	circles.clear();
	bias.clear();
	cv::Mat oimage = image.clone();
	float scalefactor = (float)setImgFixWidth / image.cols;
	resize(image, image, cv::Size(cvRound(scalefactor*image.cols), cvRound(scalefactor*image.rows)));
	GaussianBlur(image, image, cv::Size(5, 5), 3);
	cv::Mat grey;
	if (image.type() == CV_8UC3)
		cvtColor(image, grey, CV_BGR2GRAY);
	else
		grey = image;
	cv::Mat thresholded;
	int _thresParam1 = 7;
	int _thresParam2 = 7;
	if (_thresParam1 < 3)
		_thresParam1 = 3;
	else if (((int)_thresParam1) % 2 != 1)
		_thresParam1 = (int)(_thresParam1 + 1);
	adaptiveThreshold(grey, thresholded, 255, cv::ADAPTIVE_THRESH_MEAN_C, cv::THRESH_BINARY_INV, _thresParam1, _thresParam2);

	HoughCircles(thresholded, circles, CV_HOUGH_GRADIENT, 1, thresholded.rows / 8);
	cv::Point icenter = cv::Point(cvRound(oimage.cols / 2), cvRound(oimage.rows / 2));
	for (size_t i = 0; i < circles.size(); i++)
	{
		circles[i][0] /= scalefactor;
		circles[i][1] /= scalefactor;
		circles[i][2] /= scalefactor;
		cv::Point cenbias = cv::Point(cvRound(circles[i][0] - icenter.x), cvRound(circles[i][1] - icenter.y));
		bias.push_back(cenbias);
	}
	return (int)bias.size();
}