解析ROSbag图片,点云。

在网上看了一些处理ROSbag文章，简单总结一下放在下面。写于2022.06.06，有点6啊。

第一，针对ROSbag的2D图像数据：

　　1. 可以写ros .launch直接拿到。参考ros-wiki.

　　2. 可以用ros python API解析bag拿到。 下面代码可以通过 docker pull leo345/rosbag2info 拿到镜像运行。

# ROS bag image, lidar data extraction.
# Using next lidar-frame of camera-frame as lidar data output.
import cv2
import os
import rosbag
from cv_bridge import CvBridge
bag_list = os.listdir('./rosbag')


def main():
    # Process bag one by one.
    for bag_name in bag_list:
        bag = rosbag.Bag('./rosbag/' + bag_name, "r")
        bridge = CvBridge()
        bag_data = bag.read_messages()
        count = 0
        timestamp = '0000000000'
        print('Extracting work starts from ' + os.getcwd())
        os.makedirs('./raw_data/%s/pcd' % bag_name[:-4])
        os.makedirs('./raw_data/%s/imgs' % bag_name[:-4])
        for topic, msg, t in bag_data:
            # Extract camera info.
            if topic == '/kitti/camera_color_left/image_raw':
                count += 1
                cv_image = bridge.imgmsg_to_cv2(msg, 'rgb8')
                timestamp = str(msg.header.stamp.to_sec())
                img_name = timestamp + '.jpg'
                cv2.imwrite('./raw_data/%s/imgs/' % bag_name[:-4] + img_name, cv_image)

            # Extract lidar info.
            if topic == '/kitti/velo/pointcloud':
                lidar_data_name = timestamp + '.csv'
                lidar_data_path = os.path.join('./raw_data/%s/pcd/' % bag_name[:-4] + lidar_data_name)
                with open(lidar_data_path, 'w') as f:
                    f.write(str(msg))

        print("Extracting ended, And in %s, image total frames num is: " % bag_name[:-4] + str(count))


if __name__ == '__main__':
    main()

　　当然上面代码顺便解析了lidar的数据存下来。

第二，针对ROSbag的3D点云数据：

　　1. 可以用ros python API解析bag拿到。 (如上面的code)

　　2. 比较好的一个包 bagpy( without ros env.)

尤其是第二个方法，非常方便。。how to详见GitHub.

 

That's all!