Ros-Kinetic<使用cv_bridge進行OpenCV和ROS影像處理>

1.環境準備

Ubuntu16.04ROS-kineticopencv3.3.1video-stream-opencv一個USB網路攝影機

video-stream-opencv是USB網路攝影機驅動，關於它的介紹，請看github:https://github.com/ros-drivers/video_stream_opencv 安裝video-stream-opencv

cd ~/catkin_ws/src/git clone https://github.com/ros-drivers/video_stream_opencv.gitcd ~/catkin_ws/catkin_make

測試是否安裝成功

在終端輸入：

rosrun video_stream_opencv test_video_resource.py

如果能從網路攝影機中看到影像說明安裝成功。 2.opencv格式的圖片與ROS中Image-message之間的相互轉換

ROS image訊息類型到opencv類型的轉換：

cv_image = bridge.imgmsg_to_cv2(image_message, desired_encoding="passthrough")

Opencv類型到ROS Image類型的轉換：

image_message = cv2_to_imgmsg(cv_image, encoding="passthrough")

3.目標

接下來要完成使用Opencv擷取圖片，通過ROS的publisher（web_cam）發布圖片到image_topic主題下，然後有一個新的listener（image-converter）訂閱這個topic,並在圖片中的制定位置畫出一個實心圓。

4.發行者

在video_stream_opencv這個package下有一個test_video_resource.py,這個檔案的 原始作用只是簡單的將OpenCv擷取到的圖片顯示出來，這裡我們要做的是通過這個node，將擷取到的圖片發布到image_topic下去，以下是我修改過的 test_video_resource.py：

#! /usr/bin/env python2# -*- coding: utf-8 -*-"""Copyright (c) 2015 PAL Robotics SL.Released under the BSD License.Created on 7/14/15@author: Sammy Pfeiffertest_video_resource.py containsa testing code to see if opencv can open a video streamuseful to debug if video_stream does not work"""import sysimport signalimport cv2import rospyfrom cv_bridge import CvBridge, CvBridgeErrorfrom sensor_msgs.msg import Imagedef quit(signum, frame):    print ''    print 'stop fusion'    sys.exit()if __name__ == '__main__':    signal.signal(signal.SIGINT, quit)    signal.signal(signal.SIGTERM, quit)    # 參數的長度    if len(sys.argv) < 2:        print "You must give an argument to open a video stream."        print "  It can be a number as video device, e.g.: 0 would be /dev/video0"        print "  It can be a url of a stream,        e.g.: rtsp://wowzaec2demo.streamlock.net/vod/mp4:BigBuckBunny_115k.mov"        print "  It can be a video file,             e.g.: myvideo.mkv"        exit(0)    resource = sys.argv[1]    # If we are given just a number, interpret it as a video device    if len(resource) < 3:        resource_name = "/dev/video" + resource        resource = int(resource)    else:        resource_name = resource    print "Trying to open resource: " + resource_name    cap = cv2.VideoCapture(resource)    if not cap.isOpened():        print "Error opening resource: " + str(resource)        print "Maybe opencv VideoCapture can't open it"        exit(0)    bridge = CvBridge()    publisher = rospy.Publisher('image_topic', Image, queue_size=10)    rospy.init_node('web_cam')    print "Correctly opened resource, starting to show feed."    rval, frame = cap.read()    while rval:        # 將opencv格式的圖片轉換為ROS可接受的msg        image_message = bridge.cv2_to_imgmsg(frame, encoding="bgr8")        try:            publisher.publish(image_message)        except CvBridgeError as e:            print(e)        rval, frame = cap.read()

以上代碼不做過多解釋，都比較簡單。特別指出以下代碼是為了堅挺鍵盤當使用者輸入Ctrl+c的時候節點終止

def quit(signum, frame):    print ''    print 'stop fusion'    sys.exit()if __name__ == '__main__':    signal.signal(signal.SIGINT, quit)    signal.signal(signal.SIGTERM, quit)    ...    while rval:        ...

5.訂閱者

訂閱者的作用就是上邊說的從image-topic中訂閱訊息，然後將其轉換為opencv格式的圖片，然後在圖片中畫出一個實心圓，在與發行者相同的檔案夾下建立一個web_cam_listener.py:

#!/usr/bin/env pythonimport roslibroslib.load_manifest('video_stream_opencv')import rospyimport cv2from sensor_msgs.msg import Imagefrom cv_bridge import CvBridge, CvBridgeErrorclass image_converter:    def __init__(self):        self.bridge = CvBridge()        self.image_sub = rospy.Subscriber("image_topic", Image, self.callback)    def callback(self, data):        try:            cv_image = self.bridge.imgmsg_to_cv2(data, "bgr8")        except CvBridgeError as e:            print(e)        (rows, cols, channels) = cv_image.shape        if cols > 60 and rows > 60:            cv2.circle(cv_image, (150, 150), 20, (30, 144, 255), -1)        cv2.imshow("Image window", cv_image)        cv2.waitKey(3)if __name__ == '__main__':    ic = image_converter()    rospy.init_node('image_converter')    try:        rospy.spin()    except KeyboardInterrupt:        print("Shutting down")    cv2.destroyAllWindows()

6.運行節點 啟動第一個終端輸入roscore; 啟動（publisher）第二個終端輸入rosrun video_stream_opencv test_video_resource.py 0；
得到如下輸出：

xzchuang@Vostro:~/catkin_ws$ rosrun video_stream_opencv test_video_resource.py 0Trying to open resource: /dev/video0Correctly opened resource, starting to show feed.

啟動（Subscriber）啟動第三個終端輸入：

rosrun video_stream_opencv web_cam_listener.py

這時候會在網路攝影機輸入中畫一個實心圓（橙色），然後顯示出來，如圖：
啟動一個新的終端調用rqt-graph外掛程式：rosrun rqt_graph rqt_graph