| ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
Hi,
When i give to the RGBDSLAM package a pointcloud and a rectified image (from a stereo camera) of different sizes, i got this error : "PointCloud, depth and visual image differ in size! Ignoring Data"
I need to have different sizes because i am using a voxelGrid filter on my pointcloud. How can i make the RGBDSLAM package accept a pointcloud and a rectified colored image of different sizes ?
This is not possible in RGBDSLAM. It needs 3D information for the features it finds in the visual image and therefore requires either a depth image or a point cloud with a raster layout of the size of the visual image. Maybe you could compute a depth image from your cloud before voxelizing it?
Asked: 2013-05-16 02:32:51 -0600
Seen: 267 times
Last updated: May 17 '13
ROS Answers is licensed under Creative Commons Attribution 3.0 Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
This is the downfall of the package. Instead of removing points, you could just set them to zero which won't remove the points itself. The voxel grid filter should give you inlier indexes which is the points that correspond to the down-sampled point cloud.