转载

Kinect For Windows V2开发日志七：照片合成与背景去除

上一篇里讲到了Kinect可以从环境中区分出人体来。因此可以利用这个功能，来把摄像头前的人合成进照片里，和利用Photoshop不同的是，这样合成进去的人是动态且实时的。

简单的思路

BodyIndex 用的是深度数据，只能用来判断画面中的点属不属于人体而不能用来直接显示画面， Color 图里的数据只能用来显示而没有其他功能。所以如果深度数据能和彩色数据配合的话，就能利用深度数据来识别出彩色数据中的哪些点属于人体。但是深度帧的分辨率是 512 x 424 ，而彩色帧的分辨率是 1920 x 1080 ，无法将他们对应起来。然而微软提供了一个叫 ICoordinateMapper 的类。

简单来说：将彩色图上的点转换到深度图的坐标系中->判断某点是否是人体->是的话从彩色图中取出此点，与背景替换。

代码

//这份代码累死哥了  #include <iostream> #include <opencv2/imgproc.hpp> #include <opencv2/highgui.hpp> #include <Kinect.h>  using   namespace   std; using   namespace   cv;  int main(void) {     IKinectSensor   * mySensor = nullptr;     GetDefaultKinectSensor(&mySensor);     mySensor->Open();  //************************准备好彩色图像的Reader并获取尺寸*******************************      int colorHeight = 0, colorWidth = 0;     IColorFrameSource   * myColorSource = nullptr;     IColorFrameReader   * myColorReader = nullptr;     IFrameDescription   * myDescription = nullptr;     {         mySensor->get_ColorFrameSource(&myColorSource);          myColorSource->OpenReader(&myColorReader);          myColorSource->get_FrameDescription(&myDescription);         myDescription->get_Height(&colorHeight);         myDescription->get_Width(&colorWidth);          myDescription->Release();         myColorSource->Release();     }  //************************准备好深度图像的Reader并获取尺寸*******************************      int depthHeight = 0, depthWidth = 0;     IDepthFrameSource   * myDepthSource = nullptr;     IDepthFrameReader   * myDepthReader = nullptr;     {         mySensor->get_DepthFrameSource(&myDepthSource);          myDepthSource->OpenReader(&myDepthReader);          myDepthSource->get_FrameDescription(&myDescription);         myDescription->get_Height(&depthHeight);         myDescription->get_Width(&depthWidth);          myDescription->Release();         myDepthSource->Release();     }      //************************准备好人体索引图像的Reader并获取尺寸****************************      int bodyHeight = 0, bodyWidth = 0;     IBodyIndexFrameSource   * myBodyIndexSource = nullptr;     IBodyIndexFrameReader   * myBodyIndexReader = nullptr;     {         mySensor->get_BodyIndexFrameSource(&myBodyIndexSource);          myBodyIndexSource->OpenReader(&myBodyIndexReader);          myDepthSource->get_FrameDescription(&myDescription);         myDescription->get_Height(&bodyHeight);         myDescription->get_Width(&bodyWidth);          myDescription->Release();         myBodyIndexSource->Release();     }  //************************为各种图像准备buffer，并且开启Mapper*****************************      UINT    colorDataSize = colorHeight * colorWidth;     UINT    depthDataSize = depthHeight * depthWidth;     UINT    bodyDataSize = bodyHeight * bodyWidth;     Mat temp = imread("test.jpg"),background;               //获取背景图     resize(temp,background,Size(colorWidth,colorHeight));   //调整至彩色图像的大小      ICoordinateMapper   * myMaper = nullptr;                //开启mapper     mySensor->get_CoordinateMapper(&myMaper);      Mat colorData(colorHeight, colorWidth, CV_8UC4);        //准备buffer     UINT16  * depthData = new UINT16[depthDataSize];     BYTE    * bodyData = new BYTE[bodyDataSize];     DepthSpacePoint * output = new DepthSpacePoint[colorDataSize];  //************************把各种图像读进buffer里，然后进行处理*****************************      while (1)     {         IColorFrame * myColorFrame = nullptr;         while (myColorReader->AcquireLatestFrame(&myColorFrame) != S_OK);   //读取color图         myColorFrame->CopyConvertedFrameDataToArray(colorDataSize * 4, colorData.data, ColorImageFormat_Bgra);         myColorFrame->Release();          IDepthFrame * myDepthframe = nullptr;         while (myDepthReader->AcquireLatestFrame(&myDepthframe) != S_OK);   //读取depth图         myDepthframe->CopyFrameDataToArray(depthDataSize, depthData);         myDepthframe->Release();          IBodyIndexFrame * myBodyIndexFrame = nullptr;                       //读取BodyIndex图         while (myBodyIndexReader->AcquireLatestFrame(&myBodyIndexFrame) != S_OK);         myBodyIndexFrame->CopyFrameDataToArray(bodyDataSize, bodyData);         myBodyIndexFrame->Release();          Mat copy = background.clone();                  //复制一份背景图来做处理         if (myMaper->MapColorFrameToDepthSpace(depthDataSize, depthData, colorDataSize, output) == S_OK)         {             for (int i = 0; i < colorHeight; ++ i)                 for (int j = 0; j < colorWidth;++ j)                 {                     DepthSpacePoint tPoint = output[i * colorWidth + j];    //取得彩色图像上的一点，此点包含了它对应到深度图上的坐标                     if (tPoint.X >= 0 && tPoint.X < depthWidth && tPoint.Y >= 0 && tPoint.Y < depthHeight)  //判断是否合法                     {                         int index = (int)tPoint.Y * depthWidth + (int)tPoint.X; //取得彩色图上那点对应在BodyIndex里的值(注意要强转)                         if (bodyData[index] <= 5)                   //如果判断出彩色图上某点是人体，就用它来替换背景图上对应的点                         {                             Vec4b   color = colorData.at<Vec4b>(i, j);                             copy.at<Vec3b>(i, j) = Vec3b(color[0], color[1], color[2]);                         }                     }                 }             imshow("TEST",copy);         }         if (waitKey(30) == VK_ESCAPE)             break;     }     delete[] depthData;         //记得各种释放     delete[] bodyData;     delete[] output;       myMaper->Release();     myColorReader->Release();     myDepthReader->Release();     myBodyIndexReader->Release();     mySensor->Close();     mySensor->Release();      return  0; }