Question

我正在尝试使用带有以下代码的kinect来检测面部。当我使用计算机的相机时，OpenCV的detectMultiScale（）会正确检测到面部，但它似乎不能处理来自kinect的图像。

Device device;
VideoFrameRef colorFrame;
VideoStream ColorStream;
Mat colorImage, grayScaleFrame;

colorStream.create(device, SENSOR_COLOR );
colorStream.start();


while (1){
    colorStream.readFrame(&colorFrame);
    const RGB888Pixel *imageBuffer = (const RGB888Pixel*)colorFrame.getData();
    colorImage.create(colorFrame.getHeight(), colorFrame.getWidth(), CV_8UC3);
    memcpy(colorImage.data, imageBuffer, 3 * colorFrame.getHeight() * colorFrame.getWidth() * sizeof(uint8_t));

    //face detection
    blur(colorImage, colorImage, Size(2, 2), Point(-1, -1), BORDER_DEFAULT);
    cvtColor(colorImage, grayScaleFrame, CV_BGR2GRAY);
    equalizeHist(grayScaleFrame, grayScaleFrame);
    classifier.detectMultiScale(grayScaleFrame, faces, 1.1, 3, CV_HAAR_FIND_BIGGEST_OBJECT|CV_HAAR_SCALE_IMAGE,  Size(30, 30) );
}

为什么会发生这种情况？

谢谢！

Answer 1

请确保按照以下步骤使用Kinect / OpenNI（1.5.4.0 +）抓取框架：

cv::Mat rgb_image, depth_map;
cv::VideoCapture device.open(CV_CAP_OPENNI);
//set RGB-Depth mapping
device.set(CV_CAP_OPENNI_DEPTH_GENERATOR_REGISTRATION, 1.0);

while(1)
{
  device.grab();
  device.retrieve(rgb_image, CV_CAP_OPENNI_BGR_IMAGE);
  device.retrieve(depth_map, CV_CAP_OPENNI_DEPTH_MAP);
}

OpenCV没有检测到Kinect的面孔

1 个答案: