基于C++实现kinect+opencv 获取深度及彩色数据

时间:2021-12-19 08:10:27

开发环境 vs2010+OPENCV2.4.10

首先,下载最新的Kinect 2 SDK  http://www.microsoft.com/en-us/kinectforwindows/develop/downloads-docs.aspx

下载之后不要插入Kinect,最好也不用插入除了键盘鼠标以外的其它USB设备,然后安装SDK,安装完成之后插入Kinect,会有安装新设备的提示。安装完成之后可以去“开始”那里找到两个新安装的软件,一个是可以显示Kinect深度图,另外一个软件展示SDK中的各种例子程序。

进入SDK的安装目录,可以找到sample这个文件夹,里面是四种语言编写的例子,其中native是C++的,managed是C#的,还有另外两种语言不熟悉,我就熟悉C++,反正只是试试的,就用C++了。

opencv+kinect .cpp

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
#include <opencv2\opencv.hpp>
#include<iostream>
//windows的头文件,必须要,不然NuiApi.h用不了
#include <Windows.h>
//Kinect for windows 的头文件
#include "NuiApi.h"
 
using namespace std;
using namespace cv;
 
#include <d3d11.h>
 
 
//最远距离(mm)
const int MAX_DISTANCE = 3500;
//最近距离(mm)
const int MIN_DISTANCE = 200;
 
const LONG m_depthWidth = 640;
const LONG m_depthHeight = 480;
const LONG m_colorWidth = 640;
const LONG m_colorHeight = 480;
const LONG cBytesPerPixel = 4;
 
int main()
{
  //彩色图像
  Mat image_rgb;
  //深度图像
  Mat image_depth;
 
  //创建一个MAT
  image_rgb.create(480,640,CV_8UC3);
  image_depth.create(480,640,CV_8UC1);
 
  //一个KINECT实例指针
  INuiSensor* m_pNuiSensor = NULL;
 
  if (m_pNuiSensor != NULL)
  {
    return 0;
  }
 
  //记录当前连接KINECT的数量(为多连接做准备)
  int iSensorCount;
  //获得当前KINECT的数量
  HRESULT hr = NuiGetSensorCount(&iSensorCount);
 
 
  //按照序列初始化KINETC实例,这里就连接了一个KINECT,所以没有用到循环
  hr = NuiCreateSensorByIndex(iSensorCount - 1, &m_pNuiSensor);
  //初始化,让其可以接收彩色和深度数据流
  hr = m_pNuiSensor->NuiInitialize(NUI_INITIALIZE_FLAG_USES_COLOR | NUI_INITIALIZE_FLAG_USES_DEPTH);
 
  //判断是否出错
  if (FAILED(hr))
  {
    cout<<"NuiInitialize failed"<<endl;
    return hr;
  }
 
  //彩色图像获取下一帧事件
  HANDLE nextColorFrameEvent = CreateEvent(NULL, TRUE, FALSE, NULL);
  //彩色图像事件句柄
  HANDLE colorStreamHandle = NULL;
  //深度图像获取下一帧事件
  HANDLE nextDepthFrameEvent = CreateEvent(NULL, TRUE, FALSE, NULL);
  //深度图像事件句柄
  HANDLE depthStreamHandle = NULL;
 
  //实例打开数据流,这里NUI_IMAGE_TYPE_COLOR表示彩色图像
  hr = m_pNuiSensor->NuiImageStreamOpen(NUI_IMAGE_TYPE_COLOR, NUI_IMAGE_RESOLUTION_640x480, 0,2,nextColorFrameEvent,&colorStreamHandle);
 
  if( FAILED( hr ) )//判断是否提取正确
  {
    cout<<"Could not open color image stream video"<<endl;
    m_pNuiSensor->NuiShutdown();
    return hr;
  }
 
  //实例打开数据流,这里NUI_IMAGE_TYPE_DEPTH表示深度图像
  hr = m_pNuiSensor->NuiImageStreamOpen(NUI_IMAGE_TYPE_DEPTH, NUI_IMAGE_RESOLUTION_640x480, 0,2, nextDepthFrameEvent, &depthStreamHandle);
 
  if( FAILED( hr ) )//判断是否提取正确
  {
    cout<<"Could not open color image stream video"<<endl;
    m_pNuiSensor->NuiShutdown();
    return hr;
  }
 
 
 
  cv::namedWindow("depth", CV_WINDOW_AUTOSIZE);
  moveWindow("depth",300,600);
  cv::namedWindow("colorImage",CV_WINDOW_AUTOSIZE);
  moveWindow("colorImage",0,200);
 
  while (1)
  {
    NUI_IMAGE_FRAME pImageFrame_rgb;
    NUI_IMAGE_FRAME pImageFrame_depth;
 
    //无限等待新的彩色数据,等到后返回
    if (WaitForSingleObject(nextColorFrameEvent, 0) == 0)
    {
      //从刚才打开数据流的流句柄中得到该帧数据,读取到的数据地址存于pImageFrame
      hr = m_pNuiSensor->NuiImageStreamGetNextFrame(colorStreamHandle, 0, &pImageFrame_rgb);
      if (FAILED(hr))
      {
        cout<<"Could not get color image"<<endl;
        m_pNuiSensor->NuiShutdown();
        return -1;
      }
 
      INuiFrameTexture *pTexture = pImageFrame_rgb.pFrameTexture;
      NUI_LOCKED_RECT lockedRect;
 
      //提取数据帧到LockedRect,它包括两个数据对象:pitch每行字节数,pBits第一个字节地址
      //并锁定数据,这样当我们读数据的时候,kinect就不会去修改它
 
 
      pTexture->LockRect(0, &lockedRect, NULL, 0);
      //确认获得的数据是否有效
      if (lockedRect.Pitch != 0)
      {
        //将数据转换为OpenCV的Mat格式
        for (int i = 0; i < image_rgb.rows; i++)
        {
          //第i行的指针
          uchar *prt = image_rgb.ptr(i);
 
          //每个字节代表一个颜色信息,直接使用uchar
          uchar *pBuffer = (uchar*)(lockedRect.pBits) + i * lockedRect.Pitch;
 
          for (int j = 0; j < image_rgb.cols; j++)
          
            prt[3 * j] = pBuffer[4 * j];//内部数据是4个字节,0-1-2是BGR,第4个现在未使用
            prt[3 * j + 1] = pBuffer[4 * j + 1];
            prt[3 * j + 2] = pBuffer[4 * j + 2];
          }
        }
 
        imshow("colorImage",image_rgb);
        //解除锁定
        pTexture->UnlockRect(0);
        //释放帧
        m_pNuiSensor->NuiImageStreamReleaseFrame(colorStreamHandle, &pImageFrame_rgb );
      }
      else
      {
        cout<<"Buffer length of received texture is bogus\r\n"<<endl;
      }
 
      BOOL nearMode;
      INuiFrameTexture* pColorToDepthTexture;
 
 
      //深度图像的处理
      if (WaitForSingleObject(nextDepthFrameEvent, INFINITE) == 0)
      {
 
        hr = m_pNuiSensor->NuiImageStreamGetNextFrame(depthStreamHandle, 0 , &pImageFrame_depth);
 
        if (FAILED(hr))
        {
          cout<<"Could not get color image"<<endl;
          NuiShutdown();
          return -1;
        }
 
        hr = m_pNuiSensor->NuiImageFrameGetDepthImagePixelFrameTexture(
          depthStreamHandle, &pImageFrame_depth, &nearMode, &pColorToDepthTexture);
        INuiFrameTexture *pTexture = pImageFrame_depth.pFrameTexture;
        NUI_LOCKED_RECT lockedRect;
        NUI_LOCKED_RECT ColorToDepthLockRect;
 
        pTexture->LockRect(0, &lockedRect, NULL, 0);
        pColorToDepthTexture->LockRect(0,&ColorToDepthLockRect,NULL,0);
 
        //归一化
        for (int i = 0; i < image_depth.rows; i++)
        {
          uchar *prt = image_depth.ptr<uchar>(i);
 
          uchar* pBuffer = (uchar*)(lockedRect.pBits) + i * lockedRect.Pitch;
          //这里需要转换,因为每个深度数据是2个字节,应将BYTE转成USHORT
          USHORT *pBufferRun = (USHORT*)pBuffer;
 
          for (int j = 0; j < image_depth.cols; j++)
          {
            //先向,将数据归一化处理,对深度距离在300mm-3500mm范围内的像素,映射到【0—255】内,
            //超出范围的,都去做是边缘像素
            if (pBufferRun[j] << 3 > MAX_DISTANCE) prt[j] = 255;
            else if(pBufferRun[j] << 3 < MIN_DISTANCE) prt[j] = 0;
            else prt[j] = (BYTE)(256 * (pBufferRun[j] << 3)/ MAX_DISTANCE);
          }
        }
        imshow("depth", image_depth);
 
 
 
        //接下来是对齐部分,将前景抠出来
 
        //存放深度点的参数
        NUI_DEPTH_IMAGE_POINT* depthPoints = new NUI_DEPTH_IMAGE_POINT[640 * 480];
        if (ColorToDepthLockRect.Pitch != 0)
        {
          HRESULT hrState = S_OK;
           
          //一个能在不同空间坐标转变的类(包括:深度,彩色,骨骼)
          INuiCoordinateMapper* pMapper;
 
          //设置KINECT实例的空间坐标系
          hrState = m_pNuiSensor->NuiGetCoordinateMapper(&pMapper);
 
          if (FAILED(hrState))
          {
            return hrState;
          }
 
          //重要的一步:从颜色空间映射到深度空间。参数说明:
          //【参数1】:彩色图像的类型
          //【参数2】:彩色图像的分辨率
          //【参数3】:深度图像的分辨率
          //【参数4】:深度图像的个数
          //【参数5】:深度像素点数
          //【参数6】:取内存的大小,个数。类型为NUI_DEPTH_IMAGE_PIXEL
          //【参数7】:存放映射结果点的参数
          hrState = pMapper->MapColorFrameToDepthFrame(NUI_IMAGE_TYPE_COLOR, NUI_IMAGE_RESOLUTION_640x480, NUI_IMAGE_RESOLUTION_640x480,
            640 * 480, (NUI_DEPTH_IMAGE_PIXEL*)ColorToDepthLockRect.pBits,640 * 480, depthPoints);
 
          if (FAILED(hrState))
          {
            return hrState;
          }
 
 
          //显示的图像
          Mat show;
          show.create(480,640,CV_8UC3);
          show = 0;
 
          for (int i = 0; i < image_rgb.rows; i++)
          {
            for (int j = 0; j < image_rgb.cols; j++)
            {
              uchar *prt_rgb = image_rgb.ptr(i);
              uchar *prt_show = show.ptr(i);
              //在内存中偏移量
              long index = i * 640 + j;
              //从保存了映射坐标的数组中获取点
              NUI_DEPTH_IMAGE_POINT depthPointAtIndex = depthPoints[index];
 
              //边界判断
              if (depthPointAtIndex.x >= 0 && depthPointAtIndex.x < image_depth.cols &&
                depthPointAtIndex.y >=0 && depthPointAtIndex.y < image_depth.rows)
              {
                //深度判断,在MIN_DISTANCE与MAX_DISTANCE之间的当成前景,显示出来
                //这个使用也很重要,当使用真正的深度像素点再在深度图像中获取深度值来判断的时候,会出错
                if (depthPointAtIndex.depth >= MIN_DISTANCE && depthPointAtIndex.depth <= MAX_DISTANCE)
                {
                  prt_show[3 * j]   = prt_rgb[j * 3];
                  prt_show[3 * j + 1] = prt_rgb[j * 3 + 1];
                  prt_show[3 * j + 2] = prt_rgb[j * 3 + 2];
                }
              }
            }
          }
          imshow("show", show);
        }
 
        delete []depthPoints;
         
        pTexture->UnlockRect(0);
        m_pNuiSensor->NuiImageStreamReleaseFrame(depthStreamHandle, &pImageFrame_depth);
      }
 
      else
      {
        cout<<"Buffer length of received texture is bogus\r\n"<<endl;
      }
    }
 
    if (cvWaitKey(20) == 27)
      break;
  }
  return 0;
}