Ubuntu16.04，kinetic
本文示例修改使用rosbag中的frame id
在http://wiki.ros.org/bag_tools中，能找到
change_camera_info.py
usage: change_frame_id.py [-h] -o OUTPUT_BAGFILE -i INPUT_BAGFILE -f FRAME_ID
-t TOPIC [TOPIC ...] reate a new bagfile from an existing one replacing the frame id of requested
topics. optional arguments:
-h, --help show this help message and exit
-o OUTPUT_BAGFILE output bagfile
-i INPUT_BAGFILE input bagfile
-f FRAME_ID desired frame_id name in the topics
-t TOPIC [TOPIC ...] topic(s) to change
准备修改包的frame id，问题是，发现完整版的ros似乎没有安装bag_tools？什么回事呢？
看了F&A后，发现是一个老生常谈的问题How to use bag_tools? bag_tools因为缺失需要c++编译的executable，这个包已经损坏了，并且只能从源码source安装，使用git clone或者直接下载整个.zip文件后，解压，将bag_tools复制到catkin_ws中，进行catkin_make，出现另外一个问题（我觉得有可能是opencv版本兼容问题，也有可能不是）：
00:04:39.904 In file included from /opt/ros/kinetic/include/opencv-3.3.1/opencv2/core.hpp:59:0,
00:04:39.904 from /opt/ros/kinetic/include/opencv-3.3.1/opencv2/core/core.hpp:48,
00:04:39.904 from /opt/ros/kinetic/include/image_proc/processor.h:37,
00:04:39.904 from /opt/ros/kinetic/include/stereo_image_proc/processor.h:37,
00:04:39.904 from /tmp/binarydeb/ros-kinetic-bag-tools-0.0.3/src/process_stereo.cpp:34:
00:04:39.904 /opt/ros/kinetic/include/opencv-3.3.1/opencv2/core/mat.hpp: In instantiation of ‘class cv::Mat_<unsigned int>’:
00:04:39.904 /opt/ros/kinetic/include/stereo_image_proc/processor.h:180:30: required from here
00:04:39.904 /opt/ros/kinetic/include/opencv-3.3.1/opencv2/core/mat.hpp:2150:50: error: no type named ‘channel_type’ in ‘class cv::DataType<unsigned int>’
00:04:39.904 typedef typename DataType<_Tp>::channel_type channel_type;
00:04:39.904
解决方法为https://github.com/srv/srv_tools/issues/17中提到的
在bag_tools/src下的三个文件extract_images.cpp, extract_stereo_images.cpp 和 process_stereo.cpp前面加上
#define OPENCV_TRAITS_ENABLE_DEPRECATED
编译通过，perfect！
接下来修改frame id，将你所要的修改frame id的rosbag文件copy到bag_tools/dataset中，
执行
./scripts/change_frame_id.py -o ./dataset/subsetcopy01.bag -i ./dataset/subset01.bag -f /openni_rgb_optical_frame -t /camera/color/image_raw
/camera/aligned_depth_to_color/image_raw
使用ls命令时，发现没有修改过的rosbag显示为绿色，修改后的则为白色。
好像ls后显示的颜色并不影响，直接忽略。