ubuntu20.04+ROS noetic在线运行单USB双目ORB_SLAM

双目摄像头主要有以下几种，各有优缺点。

• 1.单USB插口，左右图像单独输出
• 2.双USB插口，左右图像单独输出（可能存在同步性问题）
• 3.双USB插口，左右图像合成输出
• 4.单USB插口，左右图像合成输出

官方版本的ORB SLAM2编译运行参考之前记录的博客

虽然在ubuntu22.04上编译和运行的，但我后来在ubuntu20.04上编译和运行，报错也都差不多，主要是OpenCV的版本问题，由于需要使用ROS在线运行，不建议使用OpenCV3，直接先安装ROS noetic，其自带OpenCV4.2.0版本，可不用自己再编译安装。

一、相机话题拆分

我的双目相机是单USB合成图像，然而ORM SLAM2双目ROS在订阅相机话题时，订阅的是左右图像两个节点，因此需要对USB相机话题进行拆封。

参考：
1. ROS调用USB双目摄像头模组
2. ROS&OpenCV下单目和双目摄像头的标定与使用

1. ROS调用自己的双目USB相机

安装usb_cam包

sudo apt install ros-noetic-usb-cam*

查看摄像头占用usb串口号(插上USB查看一次，拔掉USB再查看一次，可确定串口号)

ls /dev/video*

启动launch文件

cd /opt/ros/noetic/share/usb_cam/launch/
sudo gedit usb_cam-test.launch 

修改如上红框几个地方，主要有usb串口号、摄像头分辨率，以及摄像头的像素格式。默认分辨率是640x480，默认像素格式是yuyv，如果不修改的的话可能显示是花的，根据自己的相机修改即可。另外，同一个串口在关机重启后可能会发生变化，如果不显示，查询以后更改即可。

打开双目摄像头

roslaunch usb_cam usb_cam-test.launch

查看topic

rostopic list

相机只有一个/usb_cam/image_raw的话题

2. 分割双目相机图像，拆分rostopic

主要思路就是首先启动usb相机，然后新建camera_split节点，该节点订阅usb_cam/image_raw，然后分割双目相机图像，发布左图像和右图像两个节点。

创建工作空间并初始化(个人习惯放在Documents文件夹下)

mkdir -p catkin_ws/src 
cd catkin_ws 
catkin_make

进入src创建ROS包并添加依赖

cd src
catkin_create_pkg camera_split cv_bridge image_transport roscpp sensor_msgs std_msgs camera_info_manager

修改camera_split包的CMakeLists.txt文件,修改include_directories

find_package(OpenCV 4.2.0 REQUIRED)
#修改include_directories:
include_directories (
    ${catkin_INCLUDE_DIRS}
    ${OpenCV_INCLUDE_DIRS}
)
#添加可执行文件
add_executable(camera_split_node src/camera_split.cpp )
#指定链接库
target_link_libraries(camera_split_node
    ${catkin_LIBRARIES}
    ${OpenCV_LIBRARIES}
)

创建源代码文件camera_split.cpp

#include 
#include 
#include 
#include 
#include 
#include 

#include 
//#include 
//#include 

using namespace std;

class CameraSplitter
{
public:
    CameraSplitter():nh_("~"),it_(nh_)
    {
        image_sub_ = it_.subscribe("/usb_cam/image_raw", 1, &CameraSplitter::imageCallback, this);
        image_pub_left_ = it_.advertiseCamera("/left_cam/image_raw", 1);
        image_pub_right_ = it_.advertiseCamera("/right_cam/image_raw", 1);
        cinfo_ =boost::shared_ptr(new camera_info_manager::CameraInfoManager(nh_));

        //读取参数服务器参数，得到左右相机参数文件的位置
        string left_cal_file = nh_.param("left_cam_file", "");
        string right_cal_file = nh_.param("right_cam_file", "");
        if(!left_cal_file.empty())
        {
            if(cinfo_->validateURL(left_cal_file)) {
                cout<<"Load left camera info file: "<loadCameraInfo(left_cal_file);
                ci_left_ = sensor_msgs::CameraInfoPtr(new sensor_msgs::CameraInfo(cinfo_->getCameraInfo()));
            }
            else {
                cout<<"Can't load left camera info file: "<validateURL(right_cal_file)) {
                cout<<"Load right camera info file: "<loadCameraInfo(right_cal_file);
                ci_right_ = sensor_msgs::CameraInfoPtr(new sensor_msgs::CameraInfo(cinfo_->getCameraInfo()));
            }
            else {
                cout<<"Can't load right camera info file: "<image(cv::Rect(0,0,cv_ptr->image.cols/2, cv_ptr->image.rows));
        rightImgROI_=cv_ptr->image(cv::Rect(cv_ptr->image.cols/2,0, cv_ptr->image.cols/2, cv_ptr->image.rows ));
        //创建两个CvImage, 用于存放原始图像的左右部分。CvImage创建时是对Mat进行引用的,不会进行数据拷贝
        leftImgPtr_=cv_bridge::CvImagePtr(new cv_bridge::CvImage(cv_ptr->header, cv_ptr->encoding,leftImgROI_) );
        rightImgPtr_=cv_bridge::CvImagePtr(new cv_bridge::CvImage(cv_ptr->header, cv_ptr->encoding,rightImgROI_) );

        //发布到/left_cam/image_raw和/right_cam/image_raw
        ci_left_->header = cv_ptr->header; 	//很重要，不然会提示不同步导致无法去畸变
        ci_right_->header = cv_ptr->header;
        sensor_msgs::ImagePtr leftPtr =leftImgPtr_->toImageMsg();
        sensor_msgs::ImagePtr rightPtr =rightImgPtr_->toImageMsg();
        leftPtr->header=msg->header; 		//很重要，不然输出的图象没有时间戳
        rightPtr->header=msg->header;
        image_pub_left_.publish(leftPtr,ci_left_);
        image_pub_right_.publish(rightPtr,ci_right_);
    }

private:
    ros::NodeHandle nh_;
    image_transport::ImageTransport it_;
    image_transport::Subscriber image_sub_;
    image_transport::CameraPublisher image_pub_left_;
    image_transport::CameraPublisher image_pub_right_;
    boost::shared_ptr cinfo_;
    sensor_msgs::CameraInfoPtr ci_left_;
    sensor_msgs::CameraInfoPtr ci_right_;

    cv::Mat leftImgROI_;
    cv::Mat rightImgROI_;
    cv_bridge::CvImagePtr leftImgPtr_=NULL;
    cv_bridge::CvImagePtr rightImgPtr_=NULL;
};

int main(int argc,char** argv)
{
    ros::init(argc,argv, "camera_split");
    CameraSplitter cameraSplitter;
    ros::spin();
    return 0;
}

创建launch文件

<launch>
    <node pkg="camera_split" type="camera_split_node" name="camera_split_node" output="screen" />
    <node pkg="image_view" type="image_view" name="image_view_left" respawn="false" output="screen">
        <remap from="image" to="/left_cam/image_raw"/>
        <param name="autosize" value="true" />
    </node>
    <node pkg="image_view" type="image_view" name="image_view_right" respawn="false" output="screen">
        <remap from="image" to="/right_cam/image_raw"/>
        <param name="autosize" value="true" />
    </node>
</launch>

运行（运行之前先启动usb_cam）

cd catkin_ws
catkin_make
source ./devel/setup.bash
roslaunch camera_split camera_split_no_calibration.launch 

二、创建双目相机参数文件

1. 棋盘格图像获取

拆分左右相机图像，按空格键捕获

• main.cpp

#include
#include
#include
#include
#include
#include
#include
#include

using namespace std;
using namespace cv;

//双目摄像头支持2560x720, 1280x480，640x240
#define FRAME_WIDTH    2560
#define FRAME_HEIGHT   960

const char* keys =
        {
                "{help h usage ? | | print this message}"
                "{@video | | Video file, if not defined try to use webcamera}"
        };

int main(int argc, char** argv)
{
    CommandLineParser parser(argc, argv, keys);
    parser.about("Video Capture");

    if (parser.has("help"))
    {
        parser.printMessage();
        return 0;
    }

    String videoFile = parser.get(0);
    if (!parser.check())
    {
        parser.printErrors();
        return 0;
    }

    VideoCapture cap;
    if (videoFile != "")
    {
        cap.open(videoFile);
    }
    else
    {
        cap.open(0);  //0-笔记本自带摄像头，1-外接usb双目摄像头
        cap.set(CV_CAP_PROP_FRAME_WIDTH, FRAME_WIDTH);  //设置捕获视频的宽度
        cap.set(CV_CAP_PROP_FRAME_HEIGHT, FRAME_HEIGHT);  //设置捕获视频的高度
    }

    if (!cap.isOpened())
    {
        cout << "摄像头打开失败!" << endl;
        return -1;
    }

    Mat frame, frame_L, frame_R;
    cap >> frame;         //从相机捕获一帧
    Mat grayImage;

    double fScale = 1.;
    Size dsize = Size(frame.cols*fScale, frame.rows*fScale);
    Mat imagedst = Mat(dsize, CV_32S);
    resize(frame, imagedst, dsize);
    char key;
    char image_left[200];
    char image_right[200];
    int cap_count = 0;
    int count = 0;
    int count1 = 0;
    int count2 = 0;
    namedWindow("图片1", 1);
    namedWindow("图片2", 1);

    while(1)
    {
        key = waitKey(50);
        cap >> frame;
        count++;

        resize(frame, imagedst, dsize);

        frame_L = imagedst(Rect(0, 0, FRAME_WIDTH/2, FRAME_HEIGHT));
        namedWindow("Video_L", 1);
        imshow("Video_L", frame_L);

        frame_R = imagedst(Rect(FRAME_WIDTH/2, 0, FRAME_WIDTH/2, FRAME_HEIGHT));
        namedWindow("Video_R", 1);
        imshow("Video_R", frame_R);

        if (key == 27)
            break;

        if (key == 32)            //使用空格键拍照
            //if (0 == (count % 100))   //每5秒定时拍照
        {
            snprintf(image_left, sizeof(image_left), "/home/juling/Documents/CLionProjects/binocular_calibration/images3/left/left%02d.jpg", ++count1);
            imwrite(image_left, frame_L);
            imshow("图片1", frame_L);

            snprintf(image_right, sizeof(image_right), "/home/juling/Documents/CLionProjects/binocular_calibration/images3/right/right%02d.jpg", ++count2);
            imwrite(image_right, frame_R);
            imshow("图片2", frame_R);
        }

    }

    cap.release();

    return 0;
}

• CmakeLists.txt

cmake_minimum_required(VERSION 3.21)
project(binocular_calibration)

set(CMAKE_CXX_STANDARD 11)

find_package( OpenCV 3.4.12 REQUIRED )
include_directories( ${OpenCV_INCLUDE_DIRS} )
aux_source_directory(. DIR_SRCS)
#add_executable(demo ${DIR_SRCS})
add_executable(binocular_calibration main.cpp)
target_link_libraries( binocular_calibration ${OpenCV_LIBS} )

2. 双目标定

OpenCV标定

• 代码结构
  

• stereo_calibration.py

# -*- coding: utf-8 -*-

import os.path
import numpy as np
import cv2
import glob

def draw_parallel_lines(limg, rimg):
    HEIGHT = limg.shape[0]
    WIDTH = limg.shape[1]
    img = np.zeros((HEIGHT, WIDTH * 2 + 20, 3))
    img[:, :WIDTH, :] = limg
    img[:, -WIDTH:, :] = rimg
    for i in range(int(HEIGHT / 32)):
        img[i * 32, :, :] = 255
    return img

# monocular camera calibration

criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 30, 0.001)
objp = np.zeros((5 * 5, 3), np.float32)
objp[:, :2] = np.mgrid[0:5, 0:5].T.reshape(-1, 2)
objp = objp * 100  # 棋盘格方格100mm

objpoints = []
imgpoints1 = []
imgpoints2 = []
root_path ='./images2'
subfix = 'images2'
image_id = 12

# 20230828 Julyer
# 左相机imgpoints1与右相机imgpoints2的维度不一样导致报错
left_imgs = glob.glob(root_path + '/left/*.jpg')
right_imgs = glob.glob(root_path + '/right/*.jpg')
for name in left_imgs:
    img_id = name.split('left')[-1]
    left_img = cv2.imread(name)
    right_img = cv2.imread(root_path + '/right/right' + img_id)
    gray1 = cv2.cvtColor(left_img, cv2.COLOR_BGR2GRAY)
    gray2 = cv2.cvtColor(right_img, cv2.COLOR_BGR2GRAY)
    ret1, corners1 = cv2.findChessboardCorners(gray1, (5, 5), cv2.CALIB_CB_ADAPTIVE_THRESH | cv2.CALIB_CB_FILTER_QUADS)
    ret2, corners2 = cv2.findChessboardCorners(gray2, (5, 5), cv2.CALIB_CB_ADAPTIVE_THRESH | cv2.CALIB_CB_FILTER_QUADS)
    if ret1 and ret2:
        objpoints.append(objp)
        corners11 = cv2.cornerSubPix(gray1, corners1, (11, 11), (-1, -1), criteria)
        corners22 = cv2.cornerSubPix(gray2, corners2, (11, 11), (-1, -1), criteria)
        imgpoints1.append(corners11)
        imgpoints2.append(corners22)
        # img1 = cv2.drawChessboardCorners(left_img, (5, 5), corners11, ret1)
        # img2 = cv2.drawChessboardCorners(right_img, (5, 5), corners22, ret2)
        # cv2.imshow('left corners', img1)
        # cv2.imshow('right corners', img2)
        # cv2.waitKey(1)
    elif not ret1:
        print('left' + img_id + ' couldn\'t be found')
    elif not ret2:
        print('right' + img_id + ' couldn\'t be found')
ret_l, mtx_l, dist_l, rvecs_l, tvecs_l = cv2.calibrateCamera(objpoints, imgpoints1, gray1.shape[::-1], None, None)
ret_r, mtx_r, dist_r, rvecs_r, tvecs_r = cv2.calibrateCamera(objpoints, imgpoints2, gray2.shape[::-1], None, None)
print('left ret: ', ret_l)
print('right ret: ', ret_r)

# binocular camera calibration
ret, mtx_l, dist_l, mtx_r, dist_r, R, T, E, F = cv2.stereoCalibrate(objpoints, imgpoints1, imgpoints2, mtx_l, dist_l,
                                                                    mtx_r, dist_r, gray1.shape[::-1])

np.savez("./parameters for calibration_" + subfix + ".npz", ret=ret, mtx_l=mtx_l, mtx_r=mtx_r, dist_l=dist_l, dist_r=dist_r, R=R, T=T, E=E, F=F)
np.savez("./points_" + subfix + ".npz", objpoints=objpoints, imgpoints1=imgpoints1, imgpoints2=imgpoints2)

print('\nintrinsic matrix of left camera=', mtx_l)
print('\nintrinsic matrix of right camera=', mtx_r)
print('\ndistortion coefficients of left camera=', dist_l)
print('\ndistortion coefficients of right camera=', dist_r)
print('\nTransformation from left camera to right:')
print('\nR=', R)
print('\nT=', T)
print('\nReprojection Error=', ret)

# stereo rectification
R1, R2, P1, P2, Q, ROI1, ROI2 = cv2.stereoRectify(mtx_l, dist_l, mtx_r, dist_r, gray1.shape[::-1], R, T, flags=0, alpha=-1)

# undistort rectifying mapping
map1_l, map2_l = cv2.initUndistortRectifyMap(mtx_l, dist_l, R1, P1, gray1.shape[::-1], cv2.CV_16SC2)  # cv2.CV_32FC1
map1_r, map2_r = cv2.initUndistortRectifyMap(mtx_r, dist_r, R2, P2, gray2.shape[::-1], cv2.CV_16SC2)
print('\nmap1_r size', np.shape(map1_r))
print('\nmap2_r size', np.shape(map2_r))

# undistort the original image, take img#12 as an example
left_id = cv2.imread(root_path + '/left/left' + str(image_id) + '.jpg')
right_id = cv2.imread(root_path + '/right/right' + str(image_id) + '.jpg')

dst_l = cv2.remap(left_id, map1_l, map2_l, cv2.INTER_LINEAR)  # cv2.INTER_CUBIC
dst_r = cv2.remap(right_id, map1_r, map2_r, cv2.INTER_LINEAR)
cv2.imshow('map dst_r', dst_r)
cv2.waitKey(0)
print('\ndst_r size', np.shape(dst_r))
img_merge = draw_parallel_lines(dst_l, dst_r)

# cv2.imwrite('./rectify_results/left03(rectified).jpg', dst_l)
# cv2.imwrite('./rectify_results/right03(rectified).jpg', dst_r)
cv2.imwrite('rectify_results/rectify' + str(image_id) + '_' + subfix + '.jpg', img_merge)
print('\nrectification has been done successfully.')

np.savez("./rectify_results/parameters for rectification_" + subfix +".npz", R1=R1, R2=R2, P1=P1, P2=P2, Q=Q, ROI1=ROI1, ROI2=ROI2)

print('\nR1=', R1)
print('\nR2=', R2)
print('\nP1=', P1)
print('\nP2=', P2)
print('\nQ=', Q)
print('\nROI1=', ROI1)
print('\nROI2=', ROI2)

标定结果：

/usr/bin/python3.8 /home/juling/Documents/PycharmProjects/Stereo-master/rovmaker/stereo_calibration.py
left ret:  0.3898234269642049
right ret:  0.4078028378647591

intrinsic matrix of left camera= [[840.80247861   0.         667.37621909]
 [  0.         840.1220566  519.95457746]
 [  0.           0.           1.        ]]

intrinsic matrix of right camera= [[838.1562009    0.         677.06068936]
 [  0.         836.94290586 500.83733639]
 [  0.           0.           1.        ]]

distortion coefficients of left camera= [[-0.00459317  0.03249505  0.00071983  0.00213802  0.02668156]]

distortion coefficients of right camera= [[ 0.00872802 -0.01583376 -0.00164319  0.00104224  0.08360213]]

Transformation from left camera to right:

R= [[ 9.99981316e-01  4.00224781e-04 -6.09985120e-03]
 [-3.85052048e-04  9.99996830e-01  2.48836542e-03]
 [ 6.10082777e-03 -2.48597017e-03  9.99978300e-01]]

T= [[-57.64570079]
 [ -0.7422294 ]
 [  0.41023682]]

Reprojection Error= 27.596230140236862

rectification has been done successfully.

R1= [[ 0.99982475  0.01329215 -0.01318275]
 [-0.01327549  0.99991097  0.00135007]
 [ 0.01319952 -0.00117483  0.99991219]]

R2= [[ 0.9998918   0.01287432 -0.00711575]
 [-0.01288329  0.99991627 -0.0012167 ]
 [ 0.00709949  0.00130824  0.99997394]]

P1= [[838.53248123   0.         684.23641968   0.        ]
 [  0.         838.53248123 506.49901199   0.        ]
 [  0.           0.           1.           0.        ]]

P2= [[ 8.38532481e+02  0.00000000e+00  6.81501434e+02 -4.83430231e+04]
 [ 0.00000000e+00  8.38532481e+02  5.06499012e+02  0.00000000e+00]
 [ 0.00000000e+00  0.00000000e+00  1.00000000e+00  0.00000000e+00]]

Q= [[ 1.00000000e+00  0.00000000e+00  0.00000000e+00 -6.84236420e+02]
 [ 0.00000000e+00  1.00000000e+00  0.00000000e+00 -5.06499012e+02]
 [ 0.00000000e+00  0.00000000e+00  0.00000000e+00  8.38532481e+02]
 [ 0.00000000e+00  0.00000000e+00  1.73454705e-02 -4.74396078e-02]]

ROI1= (27, 13, 1221, 909)

ROI2= (31, 38, 1213, 903)

Process finished with exit code 0

3. 创建yaml参数文件

参考：https://blog.csdn.net/weixin_37918890/article/details/95626004

%YAML:1.0

#--------------------------------------------------------------------------------------------
# Camera Parameters. Adjust them!
#--------------------------------------------------------------------------------------------

# Camera calibration and distortion parameters (OpenCV) 
# Pr矩阵中的值（参考：https://blog.csdn.net/weixin_37918890/article/details/95626004）
Camera.fx: 8.38532481e+02
Camera.fy: 8.38532481e+02
Camera.cx: 6.81501434e+02
Camera.cy: 5.06499012e+02

Camera.k1: 0.0
Camera.k2: 0.0
Camera.p1: 0.0
Camera.p2: 0.0

Camera.width: 1280
Camera.height: 960

# Camera frames per second 
Camera.fps: 20.0

# stereo baseline times fx
# Pr中的值，单位转为m，取绝对值
Camera.bf: 48.3430231

# Color order of the images (0: BGR, 1: RGB. It is ignored if images are grayscale)
Camera.RGB: 1

# Close/Far threshold. Baseline times.
ThDepth: 18

#--------------------------------------------------------------------------------------------
# Stereo Rectification. Only if you need to pre-rectify the images.
# Camera.fx, .fy, etc must be the same as in LEFT.P
#--------------------------------------------------------------------------------------------
LEFT.height: 960
LEFT.width: 1280
LEFT.D: !!opencv-matrix
   rows: 1
   cols: 5
   dt: d
   data:[-0.00459317, 0.03249505, 0.00071983, 0.00213802, 0.02668156]
LEFT.K: !!opencv-matrix
   rows: 3
   cols: 3
   dt: d
   data: [840.80247861, 0., 667.37621909, 0.0, 840.1220566, 519.95457746, 0.0, 0.0, 1.0]
LEFT.R:  !!opencv-matrix
   rows: 3
   cols: 3
   dt: d
   data: [ 0.99982475,  0.01329215, -0.01318275, -0.01327549, 0.99991097, 0.00135007,  0.01319952, -0.00117483,  0.99991219]
LEFT.P:  !!opencv-matrix
   rows: 3
   cols: 4
   dt: d
   data: [838.53248123, 0. , 684.23641968, 0. , 0. , 838.53248123, 506.49901199, 0. , 0. , 0. , 1. , 0.]

RIGHT.height: 960
RIGHT.width: 1280
RIGHT.D: !!opencv-matrix
   rows: 1
   cols: 5
   dt: d
   data:[0.00872802, -0.01583376, -0.00164319, 0.00104224, 0.08360213]
RIGHT.K: !!opencv-matrix
   rows: 3
   cols: 3
   dt: d
   data: [838.1562009, 0., 677.06068936, 0.0, 836.94290586, 500.83733639, 0.0, 0.0, 1]
RIGHT.R:  !!opencv-matrix
   rows: 3
   cols: 3
   dt: d
   data: [0.9998918, 0.01287432, -0.00711575, -0.01288329, 0.99991627, -0.0012167, 0.00709949, 0.00130824, 0.99997394]
# -4.83430231e+04转为m单位，即-4.83430231e+01
RIGHT.P:  !!opencv-matrix
   rows: 3
   cols: 4
   dt: d
   data: [8.38532481e+02, 0.00000000e+00, 6.81501434e+02, -4.83430231e+01, 0, 8.38532481e+02, 5.06499012e+02, 0, 0, 0, 1, 0]

#--------------------------------------------------------------------------------------------
# ORB Parameters
#--------------------------------------------------------------------------------------------

# ORB Extractor: Number of features per image
ORBextractor.nFeatures: 1200

# ORB Extractor: Scale factor between levels in the scale pyramid 	
ORBextractor.scaleFactor: 1.2

# ORB Extractor: Number of levels in the scale pyramid	
ORBextractor.nLevels: 8

# ORB Extractor: Fast threshold
# Image is divided in a grid. At each cell FAST are extracted imposing a minimum response.
# Firstly we impose iniThFAST. If no corners are detected we impose a lower value minThFAST
# You can lower these values if your images have low contrast			
ORBextractor.iniThFAST: 20
ORBextractor.minThFAST: 7

#--------------------------------------------------------------------------------------------
# Viewer Parameters
#--------------------------------------------------------------------------------------------
Viewer.KeyFrameSize: 0.05
Viewer.KeyFrameLineWidth: 1
Viewer.GraphLineWidth: 0.9
Viewer.PointSize:2
Viewer.CameraSize: 0.08
Viewer.CameraLineWidth: 3
Viewer.ViewpointX: 0
Viewer.ViewpointY: -0.7
Viewer.ViewpointZ: -1.8
Viewer.ViewpointF: 500

三、ROS在线运行ORB SLAM2建立稀疏地图

1. 修改订阅的相机话题为拆分后的话题

复制ros_stereo.cc为ros_stereo_rovmaker.cc，修改如下部分

    ros::NodeHandle nh;

    //message_filters::Subscriber<sensor_msgs::Image> left_sub(nh, "/camera/left/image_raw", 1);
    //message_filters::Subscriber<sensor_msgs::Image> right_sub(nh, "camera/right/image_raw", 1);
    message_filters::Subscriber<sensor_msgs::Image> left_sub(nh, "/left_cam/image_raw", 1);
    message_filters::Subscriber<sensor_msgs::Image> right_sub(nh, "right_cam/image_raw", 1);
    typedef message_filters::sync_policies::ApproximateTime<sensor_msgs::Image, sensor_msgs::Image> sync_pol;
    message_filters::Synchronizer<sync_pol> sync(sync_pol(10), left_sub,right_sub);
    sync.registerCallback(boost::bind(&ImageGrabber::GrabStereo,&igb,_1,_2));

2. 重新编译

export ROS_PACKAGE_PATH=${ROS_PACKAGE_PATH}:/home/juling/Documents/projects/ORB_SLAM2_binocular
chmod +x build_ros.sh
./build_ros.sh

3. 运行

rosrun ORB_SLAM2 StereoRovmaker Vocabulary/ORBvoc.txt Examples/ROS/ORB_SLAM2/rovmaker.yaml true

四、OpenCV在线运行ORB SLAM2建立稀疏地图

参考：十里桃园的博客
由于是单usb合成图像输出，这里修改了一下代码，输出左右帧。复制Example/Stereo/stereo_euroc.cc，修改为如下代码。

• stereo_euroc_slty.cc

/**
* This file is part of ORB-SLAM2.
*
* Copyright (C) 2014-2016 Raúl Mur-Artal  (University of Zaragoza)
* For more information see 
*
* ORB-SLAM2 is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* ORB-SLAM2 is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with ORB-SLAM2. If not, see .
*/


#include
#include
#include
#include
#include

#include
#include
#include
#include

#include
using namespace std::chrono;
using namespace std;
using namespace cv;


#define FRAME_WIDTH    2560
#define FRAME_HEIGHT   960

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


    // Retrieve paths to images
    vector vstrImageLeft;
    vector vstrImageRight;
    vector vTimeStamp;
    //LoadImages(string(argv[3]), string(argv[4]), string(argv[5]), vstrImageLeft, vstrImageRight, vTimeStamp);

    //if(vstrImageLeft.empty() || vstrImageRight.empty())
   // {
      //  cerr << "ERROR: No images in provided path." << endl;
       // return 1;
    //}

   // if(vstrImageLeft.size()!=vstrImageRight.size())
   // {
     //   cerr << "ERROR: Different number of left and right images." << endl;
   //     return 1;
   // }

    // Read rectification parameters
    cv::FileStorage fsSettings(argv[2], cv::FileStorage::READ);
    if(!fsSettings.isOpened())
    {
        cerr << "ERROR: Wrong path to settings" << endl;
        return -1;
    }

    cv::Mat K_l, K_r, P_l, P_r, R_l, R_r, D_l, D_r;
    fsSettings["LEFT.K"] >> K_l;
    fsSettings["RIGHT.K"] >> K_r;

    fsSettings["LEFT.P"] >> P_l;
    fsSettings["RIGHT.P"] >> P_r;

    fsSettings["LEFT.R"] >> R_l;
    fsSettings["RIGHT.R"] >> R_r;

    fsSettings["LEFT.D"] >> D_l;
    fsSettings["RIGHT.D"] >> D_r;

    int rows_l = fsSettings["LEFT.height"];
    int cols_l = fsSettings["LEFT.width"];
    int rows_r = fsSettings["RIGHT.height"];
    int cols_r = fsSettings["RIGHT.width"];

    if(K_l.empty() || K_r.empty() || P_l.empty() || P_r.empty() || R_l.empty() || R_r.empty() || D_l.empty() || D_r.empty() ||
            rows_l==0 || rows_r==0 || cols_l==0 || cols_r==0)
    {
        cerr << "ERROR: Calibration parameters to rectify stereo are missing!" << endl;
        return -1;
    }

    cv::Mat M1l,M2l,M1r,M2r;
    cv::initUndistortRectifyMap(K_l,D_l,R_l,P_l.rowRange(0,3).colRange(0,3),cv::Size(cols_l,rows_l),CV_32F,M1l,M2l);
    cv::initUndistortRectifyMap(K_r,D_r,R_r,P_r.rowRange(0,3).colRange(0,3),cv::Size(cols_r,rows_r),CV_32F,M1r,M2r);


   // const int nImages = vstrImageLeft.size();

    // Create SLAM system. It initializes all system threads and gets ready to process frames.
    ORB_SLAM2::System SLAM(argv[1],argv[2],ORB_SLAM2::System::STEREO,true);

    // Vector for tracking time statistics
    vector vTimesTrack;
    cout << endl << "-------" << endl;
    cout << "Start processing camera ..." << endl;

  
    cv::Mat imLeft, imRight, imLeftRect, imRightRect;
//***********************************************************************8
    cv::VideoCapture cap(0, cv::CAP_V4L2);
    if (!cap.isOpened())
    {
        cout << "摄像头打开失败!" << endl;
        return -1;
    }
    else
    {
        cap.open(0, cv::CAP_V4L2);  //0-笔记本自带摄像头，1-外接usb双目摄像头
        cap.set(cv::CAP_PROP_FRAME_WIDTH, FRAME_WIDTH);  //设置捕获视频的宽度
        cap.set(cv::CAP_PROP_FRAME_HEIGHT, FRAME_HEIGHT);  //设置捕获视频的高度
        cap.set(cv::CAP_PROP_FPS, 30);
    }

    cv::Mat frame;
    cap >> frame;         //从相机捕获一帧
    cv::Mat grayImage;

    double fScale = 1.;
    cv::Size dsize = cv::Size(frame.cols*fScale, frame.rows*fScale);
    cv::Mat imagedst = cv::Mat(dsize, CV_32S);
//***********************************************************************8
	long int nImages = 0;
    int ni=0;
// Main loop
    while(ni>-1)
    {
        cap >> frame;
        cv::resize(frame, imagedst, dsize);
        imLeft = imagedst(cv::Rect(0, 0, FRAME_WIDTH/2, FRAME_HEIGHT));
        imRight = imagedst(cv::Rect(FRAME_WIDTH/2, 0, FRAME_WIDTH/2, FRAME_HEIGHT));
//***********************************************************************8
        if(imLeft.empty())
        {
            cerr << endl << "Check Left Camera!! "<< endl;
            return 1;
        }

        if(imRight.empty())
        {
            cerr << endl << "Check Right Camera!! "<< endl;
            return 1;
        }

        cv::remap(imLeft,imLeftRect,M1l,M2l,cv::INTER_LINEAR);
        cv::remap(imRight,imRightRect,M1r,M2r,cv::INTER_LINEAR);

        time_point now = system_clock::now();
        
        double tframe = now.time_since_epoch().count();
        vTimeStamp.push_back(tframe);

#ifdef COMPILEDWITHC11
        std::chrono::steady_clock::time_point t1 = std::chrono::steady_clock::now();
#else
        std::chrono::monotonic_clock::time_point t1 = std::chrono::monotonic_clock::now();
#endif

        // Pass the images to the SLAM system
        SLAM.TrackStereo(imLeftRect,imRightRect,tframe);

#ifdef COMPILEDWITHC11
        std::chrono::steady_clock::time_point t2 = std::chrono::steady_clock::now();
#else
        std::chrono::monotonic_clock::time_point t2 = std::chrono::monotonic_clock::now();
#endif

        double ttrack= std::chrono::duration_cast >(t2 - t1).count();
       
        vTimesTrack.push_back(ttrack);

        // Wait to load the next frame
/*        
	double T=0;
        if(ni0)
            T = tframe-vTimeStamp[ni-1];

       if(ttrack

• /ORB_SLAM2_binocular/CmakeLists.txt

set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${PROJECT_SOURCE_DIR}/Examples/Stereo)

add_executable(stereo_kitti
Examples/Stereo/stereo_kitti.cc)
target_link_libraries(stereo_kitti ${PROJECT_NAME})

add_executable(stereo_euroc
Examples/Stereo/stereo_euroc.cc)
target_link_libraries(stereo_euroc ${PROJECT_NAME})
# 增加下面几行
add_executable(stereo_euroc_slty
Examples/Stereo/stereo_euroc_slty.cc)
target_link_libraries(stereo_euroc_slty ${PROJECT_NAME})

重新编译

export ROS_PACKAGE_PATH=${ROS_PACKAGE_PATH}:/home/juling/Documents/projects/ORB_SLAM2_binocular
chmod +x build.sh
./build.sh

运行

./Examples/Stereo/stereo_euroc_slty Vocabulary/ORBvoc.txt Examples/ROS/ORB_SLAM2/rovmaker.yaml

yaml文件中的特征点数量ORBextractor.nFeatures从1200改成了2500，初始化的时候要慢一些，相机移动速度要平稳。
办公室稀疏建图结果：