Kinect2入门+opencv画骨架+骨架数据

最新推荐文章于 2022-03-02 19:20:01 发布
最新推荐文章于 2022-03-02 19:20:01 发布
阅读量1.5w 收藏 94
点赞数 14
分类专栏： -----语言相关----- 【___kinect___】 opencv -----体感相关----- 文章标签： Kinect2 opencv C++
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/zmdsjtu/article/details/52275879
版权

准备工作///

首先需要下载安装Kinect2的SDK，下载地址如下：

https://www.microsoft.com/en-us/download/details.aspx?id=44561

建议安装时从官网下载，之前有出现过拷贝的文件无法正常安装的情况。

 

//注意V2.0以下版本为一代Kinect所用，一代Kinect推荐V1.8

 

相对于一代的Kinect，二代由于SDK只出过这一个版本，仍有许多瑕疵，但整体性能提升了许多，手部多了几个指关节以及能同时track六个人是最大的亮点。

 

安装完SDK之后连接Kinect到USB3.0的接口测试是否正常运行。

//Windows SDK的一个优势在于安装一个SDK之后即可

 

打开SDK Browser v2.0

 

选择比较直观的Color Basics-D2D应用进行测试


/%由于二代的未连上Kinect不会像一代那样报错，这样的好处是如果接口松动也不会致使程序崩溃，但缺点也很明显，如果Kinect只是作为一个后台检测的部分便会给程序联调带来一点麻烦。%/

当然在这里没有一点麻烦，打开（点击Run按钮）之后如果能正常显示真彩图，那么基本可以说硬件部分没有问题了。

 

 

/利用C++以及opencv画出骨架

代码链接：http://download.csdn.net/detail/zmdsjtu/9609472

 

         
#include <opencv2/opencv.hpp>
#include <Windows.h>
#include <Kinect.h>
#include<iostream>
#include<time.h>

using namespace std;
using namespace cv;

//释放接口需要自己定义
template<class Interface>
inline void SafeRelease(Interface *& pInterfaceToRelease)
{
	if (pInterfaceToRelease != NULL) {
		pInterfaceToRelease->Release();
		pInterfaceToRelease = NULL;
	}	
}
 

void DrawBone(Mat& SkeletonImage, CvPoint pointSet[], const Joint* pJoints, int whichone, JointType joint0, JointType joint1);

void drawSkeleton(Mat& SkeletonImage, CvPoint pointSet[], const Joint* pJoints, int whichone);


int main(int argc, char **argv[])
{

	//OpenCV中开启CPU的硬件指令优化功能函数
	setUseOptimized(true);

	// Sensor
	IKinectSensor* pSensor;
	HRESULT hResult = S_OK;
	hResult = GetDefaultKinectSensor(&pSensor);
	hResult = pSensor->Open();
	if (FAILED(hResult)) {
		std::cerr << "Error : IKinectSensor::Open()" << std::endl;
		return -1;
	}
	//Source
	IColorFrameSource* pColorSource;
	hResult = pSensor->get_ColorFrameSource(&pColorSource);
	if (FAILED(hResult)) {
		std::cerr << "Error : IKinectSensor::get_ColorFrameSource()" << std::endl;
		return -1;
	}

	IBodyFrameSource* pBodySource;
	hResult = pSensor->get_BodyFrameSource(&pBodySource);
	if (FAILED(hResult)) {
		std::cerr << "Error : IKinectSensor::get_BodyFrameSource()" << std::endl;
		return -1;
	}

	// Reader
	IColorFrameReader* pColorReader;
	hResult = pColorSource->OpenReader(&pColorReader);
	if (FAILED(hResult)) {
		std::cerr << "Error : IColorFrameSource::OpenReader()" << std::endl;
		return -1;
	}

	IBodyFrameReader* pBodyReader;
	hResult = pBodySource->OpenReader(&pBodyReader);
	if (FAILED(hResult)) {
		std::cerr << "Error : IBodyFrameSource::OpenReader()" << std::endl;
		return -1;
	}

	// Description
	IFrameDescription* pDescription;
	hResult = pColorSource->get_FrameDescription(&pDescription);
	if (FAILED(hResult)) {
		std::cerr << "Error : IColorFrameSource::get_FrameDescription()" << std::endl;
		return -1;
	}

	int width = 0;
	int height = 0;
	pDescription->get_Width(&width); // 1920
	pDescription->get_Height(&height); // 1080
	unsigned int bufferSize = width * height * 4 * sizeof(unsigned char);

	cv::Mat bufferMat(height, width, CV_8UC4);
	cv::Mat bodyMat(height / 2, width / 2, CV_8UC4);
	cv::namedWindow("Body");

	// Color Table
	cv::Vec3b color[BODY_COUNT];
	color[0] = cv::Vec3b(255, 0, 0);
	color[1] = cv::Vec3b(0, 255, 0);
	color[2] = cv::Vec3b(0, 0, 255);
	color[3] = cv::Vec3b(255, 255, 0);
	color[4] = cv::Vec3b(255, 0, 255);
	color[5] = cv::Vec3b(0, 255, 255);

	float Skeletons[6][25][3];
	// Coordinate Mapper
	ICoordinateMapper* pCoordinateMapper;
	hResult = pSensor->get_CoordinateMapper(&pCoordinateMapper);
	if (FAILED(hResult)) {
		std::cerr << "Error : IKinectSensor::get_CoordinateMapper()" << std::endl;
		return -1;
	}


	while (1) {
		// Frame
		IColorFrame* pColorFrame = nullptr;
		hResult = pColorReader->AcquireLatestFrame(&pColorFrame);
		if (SUCCEEDED(hResult)) {
			hResult = pColorFrame->CopyConvertedFrameDataToArray(bufferSize, reinterpret_cast<BYTE*>(bufferMat.data), ColorImageFormat::ColorImageFormat_Bgra);
			if (SUCCEEDED(hResult)) {
				cv::resize(bufferMat, bodyMat, cv::Size(), 0.5, 0.5);
			}

		}

		//更新骨骼帧
		IBodyFrame* pBodyFrame = nullptr;
		hResult = pBodyReader->AcquireLatestFrame(&pBodyFrame);
		if (SUCCEEDED(hResult)) {
			IBody* pBody[BODY_COUNT] = { 0 };
			//更新骨骼数据
			hResult = pBodyFrame->GetAndRefreshBodyData(BODY_COUNT, pBody);
			if (SUCCEEDED(hResult)) {
				for (int count = 0; count < BODY_COUNT; count++) {
					BOOLEAN bTracked = false;
					hResult = pBody[count]->get_IsTracked(&bTracked);
					if (SUCCEEDED(hResult) && bTracked) {
						Joint joint[JointType::JointType_Count];
	/					
						hResult = pBody[count]->GetJoints(JointType::JointType_Count, joint);//joint
						if (SUCCEEDED(hResult)) {
							// Left Hand State
							HandState leftHandState = HandState::HandState_Unknown;
							hResult = pBody[count]->get_HandLeftState(&leftHandState);
							if (SUCCEEDED(hResult)) {
								ColorSpacePoint colorSpacePoint = { 0 };
								hResult = pCoordinateMapper->MapCameraPointToColorSpace(joint[JointType::JointType_HandLeft].Position, &colorSpacePoint);
								if (SUCCEEDED(hResult)) {
									int x = static_cast<int>(colorSpacePoint.X);
									int y = static_cast<int>(colorSpacePoint.Y);


									if ((x >= 0) && (x < width) && (y >= 0) && (y < height)) {
										if (leftHandState == HandState::HandState_Open) {
											cv::circle(bufferMat, cv::Point(x, y), 75, cv::Scalar(0, 128, 0), 5, CV_AA);

										}
										else if (leftHandState == HandState::HandState_Closed) {
											cv::circle(bufferMat, cv::Point(x, y), 75, cv::Scalar(0, 0, 128), 5, CV_AA);
										}
										else if (leftHandState == HandState::HandState_Lasso) {
											cv::circle(bufferMat, cv::Point(x, y), 75, cv::Scalar(128, 128, 0), 5, CV_AA);
										}
									}
								}
							}

						HandState rightHandState = HandState::HandState_Unknown;
                        ColorSpacePoint colorSpacePoint = { 0 };
						hResult = pBody[count]->get_HandRightState(&rightHandState);
							if (SUCCEEDED(hResult)) {
								hResult = pCoordinateMapper->MapCameraPointToColorSpace(joint[JointType::JointType_HandRight].Position, &colorSpacePoint);
								if (SUCCEEDED(hResult)) {
									int x = static_cast<int>(colorSpacePoint.X);
									int y = static_cast<int>(colorSpacePoint.Y);

									if ((x >= 0) && (x < width) && (y >= 0) && (y < height)) {
										if (rightHandState == HandState::HandState_Open) {
											cv::circle(bufferMat, cv::Point(x, y), 75, cv::Scalar(0, 128, 0), 5, CV_AA);
										}
										else if (rightHandState == HandState::HandState_Closed) {
											cv::circle(bufferMat, cv::Point(x, y), 75, cv::Scalar(0, 0, 128), 5, CV_AA);

										}
										else if (rightHandState == HandState::HandState_Lasso) {
											cv::circle(bufferMat, cv::Point(x, y), 75, cv::Scalar(128, 128, 0), 5, CV_AA);
										}
									}
								}
							}
							CvPoint skeletonPoint[BODY_COUNT][JointType_Count] = { cvPoint(0,0) };
							// Joint
							for (int type = 0; type < JointType::JointType_Count; type++) {
								ColorSpacePoint colorSpacePoint = { 0 };
								pCoordinateMapper->MapCameraPointToColorSpace(joint[type].Position, &colorSpacePoint);
								int x = static_cast<int>(colorSpacePoint.X);
								int y = static_cast<int>(colorSpacePoint.Y);
								skeletonPoint[count][type].x = x;
								skeletonPoint[count][type].y = y;
								if ((x >= 0) && (x < width) && (y >= 0) && (y < height)) {
									cv::circle(bufferMat, cv::Point(x, y), 5, static_cast< cv::Scalar >(color[count]), -1, CV_AA);
								}
							}
							for (int i = 0; i < 25; i++) {
								Skeletons[count][i][0] = joint[i].Position.X;
								Skeletons[count][i][1] = joint[i].Position.Y;
								Skeletons[count][i][2] = joint[i].Position.Z;
							}
								drawSkeleton(bufferMat, skeletonPoint[count], joint, count);			           		
						}
					}
				}
				cv::resize(bufferMat, bodyMat, cv::Size(), 0.5, 0.5);
			}
			for (int count = 0; count < BODY_COUNT; count++) {
				SafeRelease(pBody[count]);
			}
		}
		SafeRelease(pColorFrame);
		SafeRelease(pBodyFrame);

		waitKey(1);
		cv::imshow("Body", bodyMat);

	}


	SafeRelease(pColorSource);
	SafeRelease(pColorReader);
	SafeRelease(pDescription);
	SafeRelease(pBodySource);
	// done with body frame reader
	SafeRelease(pBodyReader);

	SafeRelease(pDescription);
	// done with coordinate mapper
	SafeRelease(pCoordinateMapper);

	if (pSensor) {
		pSensor->Close();
	}
	SafeRelease(pSensor);

	return 0;
}


void DrawBone(Mat& SkeletonImage, CvPoint pointSet[], const Joint* pJoints, int whichone, JointType joint0, JointType joint1)
{
	TrackingState joint0State = pJoints[joint0].TrackingState;
	TrackingState joint1State = pJoints[joint1].TrackingState;

	// If we can't find either of these joints, exit
	if ((joint0State == TrackingState_NotTracked) || (joint1State == TrackingState_NotTracked))
	{
		return;
	}

	// Don't draw if both points are inferred
	if ((joint0State == TrackingState_Inferred) && (joint1State == TrackingState_Inferred))
	{
		return;
	}


	CvScalar color;
	switch (whichone) //跟踪不同的人显示不同的颜色   
	{
	case 0:
		color = cvScalar(255);
		break;
	case 1:
		color = cvScalar(0, 255);
		break;
	case 2:
		color = cvScalar(0, 0, 255);
		break;
	case 3:
		color = cvScalar(255, 255, 0);
		break;
	case 4:
		color = cvScalar(255, 0, 255);
		break;
	case 5:
		color = cvScalar(0, 255, 255);
		break;
	} 


	// We assume all drawn bones are inferred unless BOTH joints are tracked
	if ((joint0State == TrackingState_Tracked) && (joint1State == TrackingState_Tracked))
	{
		line(SkeletonImage, pointSet[joint0], pointSet[joint1], color, 2);
	}
	else
	{
		line(SkeletonImage, pointSet[joint0], pointSet[joint1], color, 2);
	}
}

void drawSkeleton(Mat& SkeletonImage, CvPoint pointSet[], const Joint* pJoints, int whichone)
{

	// Draw the bones

	// Torso
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_Head, JointType_Neck);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_Neck, JointType_SpineShoulder);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_SpineShoulder, JointType_SpineMid);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_SpineMid, JointType_SpineBase);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_SpineShoulder, JointType_ShoulderRight);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_SpineShoulder, JointType_ShoulderLeft);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_SpineBase, JointType_HipRight);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_SpineBase, JointType_HipLeft);

	// Right Arm    
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_ShoulderRight, JointType_ElbowRight);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_ElbowRight, JointType_WristRight);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_WristRight, JointType_HandRight);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_HandRight, JointType_HandTipRight);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_WristRight, JointType_ThumbRight);

	// Left Arm
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_ShoulderLeft, JointType_ElbowLeft);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_ElbowLeft, JointType_WristLeft);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_WristLeft, JointType_HandLeft);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_HandLeft, JointType_HandTipLeft);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_WristLeft, JointType_ThumbLeft);

	// Right Leg
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_HipRight, JointType_KneeRight);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_KneeRight, JointType_AnkleRight);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_AnkleRight, JointType_FootRight);

	// Left Leg
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_HipLeft, JointType_KneeLeft);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_KneeLeft, JointType_AnkleLeft);
	DrawBone(SkeletonImage, pointSet, pJoints, whichone, JointType_AnkleLeft, JointType_FootLeft);
}



新建一个空白项目，在配置opencv环境的基础之上

包含目录要加上C:\Program Files\Microsoft SDKs\Kinect\v2.0_1409\inc

库目录要加上C:\Program Files\Microsoft SDKs\Kinect\v2.0_1409\Lib\x86

输入附加依赖项要加上kinect20.lib

之前配置opencv环境的链接：http://blog.csdn.net/zmdsjtu/article/details/52235056

 

配合上连接的代码应该可以正常运行了。

注意：因为直接VS调试的环境没有系统默认的opencv环境变量，故而可以生成之后直接点击对应EXE运行，否则会提示缺少Dll,嫌麻烦也可以把指定dll放到编译目录下

 

为方便后续程序的开发定义了：

float Skeletons[6][25][3];//六副骨架储存的数据，其他功能模块较为直观，也相对简单。


//有一点缺陷，画图的时候股价点有时会跑到（0，0）的位置，如果发现代码里的bug请评论或者私信下哈，谢啦~


运行结果如下：


 

 最后祝大家开发愉快~

 


附录：各个关节点对应数字

        JointType_SpineBase	= 0,
        JointType_SpineMid	= 1,
        JointType_Neck	= 2,
        JointType_Head	= 3,
        JointType_ShoulderLeft	= 4,
        JointType_ElbowLeft	= 5,
        JointType_WristLeft	= 6,
        JointType_HandLeft	= 7,
        JointType_ShoulderRight	= 8,
        JointType_ElbowRight	= 9,
        JointType_WristRight	= 10,
        JointType_HandRight	= 11,
        JointType_HipLeft	= 12,
        JointType_KneeLeft	= 13,
        JointType_AnkleLeft	= 14,
        JointType_FootLeft	= 15,
        JointType_HipRight	= 16,
        JointType_KneeRight	= 17,
        JointType_AnkleRight	= 18,
        JointType_FootRight	= 19,
        JointType_SpineShoulder	= 20,
        JointType_HandTipLeft	= 21,
        JointType_ThumbLeft	= 22,
        JointType_HandTipRight	= 23,
        JointType_ThumbRight	= 24,
        JointType_Count	= ( JointType_ThumbRight + 1 )



朱铭德
关注
  • 14
    点赞
  • 踩
  • 94
    收藏
    觉得还不错? 一键收藏
  • 打赏
    打赏
  • 42
    评论
专栏目录
Kinect2.0+Opencv人体骨骼捕捉及绘制VisualStudio2015工程文件
01-14
使用Kinect2.0获取人体骨骼，并使用openCV实时绘制，里面为VisualStudio2015工程文件，配置好Kinect开发环境可直接用VS2015打开并运行
谈谈NITE 2与OpenCV结合的第二个程序（提取人体骨骼坐标）
weixin_33766805的博客
01-16 256
温故而知新——NITE 2的基本使用主要包括以下几个步骤： 1. 初始化NITE环境： nite::NiTE::initialize(); 2. 创建User跟踪器： nite::UserTracker mUserTracker; mUserTracker.create(); 3. 创建并读取User Frame信息：nite::UserTrackerFram...
Kinect 2.0 + OpenCV 显示深度数据、骨架信息、手势状态和人物二值图
江饭
11-04 1万+
使用Kinect 2.0 + OpenCV 获取深度信息（Depth Data），骨架信息（Body Data），手势状态（Hand State）和人物二值图（Body Index Data）。
Kinect开发学习笔记之（七）骨骼数据的提取
热门推荐
zouxy09的专栏
11-08 10万+
Kinect开发学习笔记之（七）骨骼数据的提取 [email protected] http://blog.csdn.net/zouxy09   我的Kinect开发平台是： Win7x86 + VS2010 + Kinect for Windows SDK v1.6 + OpenCV2.3.0 开发环境的搭建见上一文： http://blog.csdn.net/zouxy09/arti
第三篇 KinectV2骨骼获取原理和获取方法及源代码
baolinq的博客
08-30 4万+
第三篇  KinectV2骨骼获取原理和获取方法及源代码   首先声明一下，本系统所使用的开发环境版本是计算机系统Windows 10、Visual Studio 2013、Opencv3.0和Kinect SDK v2.0。这些都可以在百度上找到，download下来安装一下即可。 通过上一篇文章，我们已经基本学会如何获取Kinect 的各种数据源，对于骨骼数据的获取是稍微复杂的那一个
Kinect程序设计
yangtao639999的专栏
10-18 2342
关键词：kinect   体感 kinect开发 日期：2011-10-18 作者：yangtao ---------------------------------------- 一． 简介 Kinect是Xbox360上的体感外设，可以被应用到PC机上，提供动作捕
Kinect V2开发（5）绘制骨架图
pancast的博客
11-07 2351
在OpenCV中给各个关节点连线，绘制骨架图。
Kinect开发学习笔记之（八）彩色、深度、骨骼和用户抠图结合
zouxy09的专栏
11-08 2万+
Kinect开发学习笔记之（八）彩色、深度、骨骼和用户抠图结合 [email protected] http://blog.csdn.net/zouxy09   我的Kinect开发平台是： Win7 x86 + VS2010 + Kinect for Windows SDK v1.6 + OpenCV2.3.0 开发环境的搭建见上一文： http://blog.csdn.net/zou
基于openCV绘制自定义形状骨骼线
qq_41343328的博客
03-02 3402
单一的线条缺乏视觉上的冲突感，所以希望利用同样的骨骼点信息画出更有“感觉”的骨格线。这里介绍一种实现方式，希望能给到大家一些启发。
骨络数据
沧浪之水清兮
07-15 353
#include #include //windows的头文件，必须要，不然NuiApi.h用不了 #include //Kinect for windows 的头文件 #include "NuiApi.h" /* #include #include "NuiSensor.h" #include "NuiSkeleton.h" #include #include */ using
opencv骨骼化
10-27
一个基于opencv2.4.6 的完整的细化算法源代码程序，opencv的细化算法
Kinect2+Opencv显示骨架
08-22
利用Opencv绘制骨骼，方便不熟悉Direct2D开发的童鞋使用~
opencv显示kinect深度彩色骨架数据
07-02
opencv下显示kinect深度数据，彩色数据和骨架数据
骨架提取与动作识别.rar_createjkb_kinect姿势_opencv_opencv 动作识别_opencv 骨架提取
07-15
基于opencv以及kinect的骨架提取和姿势识别代码，C++环境
kinect骨架提取到opencv窗口显示
05-24
kinect骨架提取到opencv窗口显示
Dlib提取人脸特征点（68点，opencv画图）
朱铭德的博客
12-04 7万+
Dlib+opencv 68点特征点的使用以及绘图。
Opencv3.2各个模块功能详细简介（包括与Opencv2.4的区别)
朱铭德的博客
02-08 4万+
Opencv3.2各个模块详细介绍以及与Opencv2.4的对比
如何在程序里写死一张图片（base64编码，OpenCV）
朱铭德的博客
07-10 4万+
首先准备一张图片，然后通过 地址 转成string
Dlib+opencv实时提取人脸轮廓（windows环境下//Dlib配置入门）
朱铭德的博客
09-03 4万+
Windows环境下使用Dlib进行人脸特征点提取以及轮廓勾画//示例程序 可以商用的人脸特征点检测里不错的选择
Kinect开发环境：WIN10+Kinect+for+windows+SDK2.0+OpenCV3.41+VS2017环境配置+原创
最新发布
09-20
根据引用和引用所提供的信息，您可以按照以下步骤配置您的WIN10 Kinect for Windows SDK2.0和OpenCV3.41开发环境： 1. 首先，下载并安装VS2015或VS2017开发环境。 2. 在您的工作目录下创建一个文件夹用于存放OpenCV的头文件和库文件。 3. 下载OpenCV3.41版本的库文件和头文件，并将它们复制到您创建的文件夹中。 4. 下载并安装Kinect for Windows SDK2.0。 5. 在您的项目中设置OpenCV和Kinect的引用路径。 6. 编写您的代码，并使用OpenCV和Kinect SDK2.0进行开发。 以下是几个

“相关推荐”对你有帮助么？

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论 42
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

朱铭德

五毛也是爱٩(●´৺`●)૭

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付：¥1
获取中
扫码支付

您的余额不足，请更换扫码支付或充值

打赏作者

实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值