Kinect for windows 开发入门 八：景深数据获取和使用 下

背景知识

1.  前面已经有提到，所谓直方图就是不同深度的像素分布图。

2.  Player id只有在打开SkeletonStream的时候才会产生。

SkeletonStream skeletonStream =kinectSensor.SkeletonStream;

       skeletonStream.Enable();

Playerid是动态分配的，同一个人走出视线再回来id会改变。Playerid也不是从0开始，从小到大排序的。

3.  如何计算人体的高度和宽度。如下图所示，根据Player id和深度数据，我们可以获取到人体的平均深度d。根据Kinect支持的水平角度（57度）和d，我们可以通过如下的两个直角三角形计算出人体所在切面的1/2实际宽度d
*Pi * （57度/2）/180度 得到b。根据DepthStream的分辨率中的横向像素总数和人体所占像素，可以知道人体所占切面比例r。然后b*2* r就得到了人体的宽度。

比如，人体深度为2m，分辨率为1024 * 768，人体所占横向像素平均为300，那么人体的宽度为（2m * Pi * 28.5度 / 180 度）* 2 * （300 / 1024）=0.58米。对于高度的计算也是一样的。




 

示例代码

1.  直方图

       privatevoidCreateDepthHistogram(short[]
pixelData, DepthImageFrame frame)

        {

            int[] depths =
newint[4096];

 

            DepthHistogram.Children.Clear();

 

            int depth = 0;

            int max = 0;

            for (int
i = 0; i <pixelData.Length; i++)

            {

                depth = pixelData[i] >>
DepthImageFrame.PlayerIndexBitmaskWidth;

                if (depth > 0 && depth < 4096)

                {

                   depths[depth]++;

                    max = depths[depth] >max ? depths[depth] : max;

                }

            }

 

            for (int
index = 0; index <depths.Length; index++)

            {

                if (depths[index] > 0)

               {

                    Rectangle rect =
newRectangle();

                    rect.Fill =
Brushes.DarkBlue;

                    rect.Width =DepthHistogram.ActualWidth / 4096;

                    rect.Height =DepthHistogram.ActualHeight * ((double)depths[index]/ max);

                    rect.Margin =
newThickness(1, 0, 1, 0);

                    rect.VerticalAlignment =System.Windows.VerticalAlignment.Bottom;

                   DepthHistogram.Children.Add(rect);

                }

            }

        }

2.  抠像

       privatevoid CatchPlayer(short[]
pixelData, DepthImageFrame frame)

        {

            int playId = 0;

            byte[] playerPixelData =
newbyte[frame.Width * frame.Height *4];

            for (int
i = 0; i <pixelData.Length; i++)

            {

                playId = pixelData[i] &
DepthImageFrame.PlayerIndexBitmask;

                if (playId > 0)

                {

                    ShowThePlayerId.Text =
string.Format("PlayerId:{0}", playId.ToString());

                    playerPixelData[i * 4] = (byte)0;

                    playerPixelData[i * 4 + 1]= (byte)0;

                    playerPixelData[i * 4 + 2]= (byte)255;

                }

                else

                {

                    playerPixelData[i * 4] = (byte)0;

                    playerPixelData[i * 4 + 1]= (byte)0;

                    playerPixelData[i * 4 + 2]= (byte)0;

                }

            }

 

            this.playerImageBitmap.WritePixels(this.playerImageBitmapRect,
playerPixelData, frame.Width * 4, 0);

        }

3.  计算高度宽度

       privatevoidCaculateHightAndWidth(DepthImageFrame
frame)

        {

            double avgDepth = 0.0;

            int sumDepth = 0;

            int pixelCount = 0;

            int minX = frame.Width;

            int maxX = 0;

            int minY = frame.Height;

            int maxY = 0;

 

            for (int
i = 0; i < frame.Width;i++)

            {

                for (int
j = 0; j < frame.Height;j++)

                {

                    if (playerData[i,j] <= 0)

                    {

                        continue;

                    }

 

                    pixelCount++;

                    minX = i < minX ? i :minX;

                    maxX = i > maxX ? i :maxX;

                    minY = j < minY ? j :minY;

                    maxY = j > maxY ? j :maxY;

 

                    sumDepth += playerData[i,j];

                }

            }

 

            if (pixelCount <= 0)

            {

                return;

            }

 

            int widthPixel = maxX - minX;

            int heightPixel = maxY - minY;

 

            avgDepth = sumDepth / pixelCount;

 

            double actualWidth = avgDepth * 28.5 * 3.14159 * 2 / 180;

            double actualHeight = avgDepth * 21.5 * 3.14159 * 2 /180;

            double width = actualWidth * widthPixel / frame.Width;

           double height = actualHeight * heightPixel /frame.Height;

 

            ShowThePlayerWidthAndHeight.Text =
string.Format("Player width:{0} \nPlayerHeight:{1}",width,
height);

        }

 

效果演示

1.  直方图，展示统一深度出现的次数。

如图所示，人，椅子，背景三个高峰。



2.  抠像

将playerid 不为0的像素显示为红色。



3.  深度宽度

很粗劣的用最大值减最小值计算出来的。偏差比较大。实际高度应该是1.6米。不过，我们的目的是展示如何计算。