0

我正在尝试订阅两个相机主题：“realsense_front_left/color/image_raw”主题和“realsense_front_right/color/image_raw”主题，这两个主题都是来自两个单目相机的数据。在此之后，我需要将左摄像头图像的左半部分添加到右摄像头图像的右半部分，然后发布合并后的图像。每次我尝试运行节点时，我都会收到一个 AttributeError：'Merger' 对象没有属性 'front_left_camera_image。我尝试了一些不同的东西，但我通常会遇到同样的错误。有人可以帮我吗？

# !/usr/bin/env python
import rospy
from sensor_msgs.msg import Image
from cv_bridge import CvBridge
import cv2
import os
import numpy as np
from PIL import Image as img

bridge = CvBridge()

class Merger:
    def __init__(self):
        self.front_left_camera = None
        self.front_right_camera = None
        self.bridge = CvBridge()
        # Node cycle rate (in Hz)
        self.loop_rate = rospy.Rate(1)

        # Publisher
        self.pub = rospy.Publisher('post_merged_image', Image, queue_size = 10)

        # Subscribers
        rospy.Subscriber("realsense_front_left/color/image_raw", Image, self.front_left_camera_callback)
        rospy.Subscriber("realsense_front_right/color/image_raw", Image, self.front_right_camera_callback)

    def front_left_camera_callback(self,msg):
        rospy.loginfo('Left camera Image received...')
        # Extract the left camera image
        self.front_left_camera_image = self.bridge.imgmsg_to_cv2(msg)

    def front_right_camera_callback(self,msg):
        rospy.loginfo('Right camera Image received...')
        # Extract the right camera image
        self.front_right_camera_image = self.bridge.imgmsg_to_cv2(msg)

    def overlap(self):
        if self.front_left_camera_image is not None and self.front_right_camera_image is not None:
            # Merging operation here (Merge the left half of left camera image to right half of right camera image)
            left_image_array = np.asarray(self.front_left_camera_image)
            h_l, w_l = left_image_array[0], left_image_array[1]
            right_image_array = np.asarray(self.front_right_camera_image)
            h_r, w_r = right_image_array[0], right_image_array[1]
            # Assusming both are equal in size (h_l = h_r and w_l = w_r)
            # Slice the half images
            left_half_array = left_image_array[h_l, 0:w_l/2, :]
            right_half_array = right_image_array[h_r, w_r/2:w_r, :]
            # concatenate the arrays
            merged_array = np.concatenate(left_half_array, right_half_array, axis = 1)
            # convert numpy array to image
            self.merged_image = img.fromarray(merged_array)
        else:
            pass

    def start(self):
        while not rospy.is_shutdown():
            rospy.loginfo("merging images")
            self.overlap()
            if self.merged_image is not None:
                rospy.loginfo("publishing merged image")
                self.pub.publish(bridge.cv2_to_imgmsg(self.merged_image))
            self.loop_rate.sleep()

if __name__ == '__main__':
    rospy.init_node('image_merger', anonymous=True)
    my_node = Merger()
    my_node.start()

`
4

1 回答 1

0

这是因为front_left_camera_image只在回调中分配，而不是在__init__. 因此，无法保证您的回调overlap()在主运行循环中被调用时已被调用。事实上，这种情况属实的可能性极小。我还应该指出，这也将是一个问题front_right_camera_image。

由于您已经在进行None检查，因此只需在__init__. 看起来您可能已经尝试过，但名称略有错误。

编辑：你__init__应该看起来像这样

def __init__(self):
    self.front_left_camera = None
    self.front_right_camera = None
    self.front_left_camera_image = None
    self.front_right_camera_image = None
    self.bridge = CvBridge()
    # Node cycle rate (in Hz)
    self.loop_rate = rospy.Rate(1)
    # Publisher
    self.pub = rospy.Publisher('post_merged_image', Image, queue_size = 10)

    # Subscribers
    rospy.Subscriber("realsense_front_left/color/image_raw", Image, self.front_left_camera_callback)
    rospy.Subscriber("realsense_front_right/color/image_raw", Image, self.front_right_camera_callback)
于 2022-01-27T14:40:25.367 回答