dtj's profile - activity

My goal is to compress a movie of kinect data into the smallest format possible. I'm investigating if I can use movie compression on either the raw /camera/ir/image_raw images or maybe the /camera/depth_registered/image_raw frames.

I'm using this tutorial to extract the frames from a rosbag recording. And it works great. For the next step, I have two questions:

1. Can you export image frames in png (i.e. a lossless) format for compression? I found this answer but I'd prefer not use openCV/python.

2. Is there a ros node to convert the raw image format into points. "convert_pointcloud_to_image" will convert points into images but I'm looking to use both the camera data & the image to get points & haven't found a tutorial or node for this...

I'm running Ubuntu 11.10 w/Electric & PCL 1.6, as I haven't had good luck with Openni drivers on 12.04. Looking in: /opt/ros/electric/stacks/perception_pcl/pcl_ros/bin I don't see anything that does what I would want. I assume this functionality exists somewhere (as the Openni drivers converts an inferred image into a point cloud) but I can't seem to find it.

Help please?

I didn't actually do this (what was below worked), but it's a great idea, would totally work, & this should be incorporated into ros. Is there a way I can +1 this somewhere?

This works, thank you!

Hi,

I'm processing a kinect point cloud with Electric + Ubuntu 11.10. I've recorded a .bag file with the topics /tf & /camera/depth/points. Using "rosbag play", rviz shows a beautiful point cloud.

I'm trying to convert the individual point clouds to pcd format so I can process them with PCL. Initially, I got this error. So, I followed the response to that question and ran:

rosrun tf static_transform_publisher 0 0 0 0 0 0 base_link openni_camera 100

While I re-recorded the data in hopes of making bag_to_pcd happy. Sadly, when I run

rosrun pcl_ros bag_to_pcd data6_2012_05_01.bag /camera/depth/points d6_expand/

I now get:

"Saving recorded sensor_msgs::PointCloud2 messages on topic /camera/depth/points to d6_expand/

Frame /base_link exists with parent NO_PARENT.

Segmentation fault"

How can I make bag_to_pcd happy? Should I attempt to change the bag_to_pcd code? Specifically,

pcl_ros::transformPointCloud ("/base_link", *cloud, cloud_t, tf_listener);

& if I should change it what should it be? Would

pcl_ros::transformPointCloud ("/tf", *cloud, cloud_t, tf_listener);

work? I'm a new to ROS so I don't fully understand what is going on yet. This seems to be a reoccurring issue on both the ROS questions and PCL users lists so I'd love to get fixed.

Thanks for you help!