In a project, we are creating a virtual tour of an apartment. We want to display the room dimension in that virtual image. So far we are using RICO theta V to create the virtual tour.
My plan is to use Lidar to measure the room length. One idea is, first i will create 2D point cloud and attach this information to the panoramic image(Question: dont know how can i do that? image fusion may be !). And from image i will detect the corner and from that particular corner i will find the respective points that i got from the Lidar. So from two corner i will find two respective points and will find the distance between two points based on Lidar.
- I want to know am I in the right track?
- is there any easier solution to solve this problem?
- How can i merge the image with 2d point cloud that i get from the Lidar?
For a 2D point cloud. Take your 3D point cloud and remove the ceiling of the room by filtering any points at or above the Z height of the ceiling. Next remove the floor of the room by filtering any points at or below the Z value of the floor. To be sure to get all (ceiling & floor) you may want to add a buffer to make sure you get all the ceiling or floor because not all true horizontal. Once you have 3D filter Z clipped the ceiling and floor you can just plot the 2D points and this should give you a high concentration of points of just the walls. With this 2D point set you can measure the distances to each wall using only the X and Y values. If you dont know how to filter in Z then just scatter plot in X and Y. This should still give you a higher concentration of points at the walls. The points for the floor and ceiling will give you residual noise in the scatter plot but you should still see a higher concentration of points at the walls, enough to take measurements. To project your panoramic image you will need to know how your camera lens field of view is. If you dont know this you will have to add control points to key sections of your panoramic image then calculate how much each pixel maps to fractions of degrees in both horizontal and vertical as in most case this is different. Once you have your fractions of degrees (polar projection) you can then convert your point cloud from XYZ to polar projection using the center of where you took the panoramic image. At this point your panoramic image should match your point cloud. I would suggest adding a plane (or facets) to the point clouds to reduce how complex your scene will be. Here is a link to something similar i did a few years back that may help:
Conference Paper Integrating Gigapixel Panoramas and LiDAR: Increasing Geolog...
Hope this helps.
- Badges
- Science topic
- Similar topics
- Design
- Architecture