I am facing a problem on 3D reconstruction since I am a new to this filed. I have some different views' depth map(point clouds), I want to use them to reconstruct the scene to get the effect like using the kinect fusion. Is there any paper of source code to settle this problem. Or any ideas on this problem.
PS:the point cloud is stored as a file with (x,y,z), you can check here to get the data.
Thank you very much.
As you have stated that you are new to this field, I shall attempt to keep this high level. Please do comment if there is something that is not clear.
The pipeline you refer to has three key stages:
Integration
Rendering
Pose Estimation
The Integration stage takes the unprojected points from a Depth Map (Kinect image) under the current pose and "integrates" them into a spatial data structure (a Voxel Volume such as a Signed Distance Function or a hierarchical structure like an Octree), often by maintaining per Voxel running averages.
The Rendering stage takes the inverse pose for the current frame and produces an image of the visible parts of the model currently in view. For the common volumetric representations this is achieved by Raycasting. The output of this stage provides the points of the model to which the next live frame is registered (the next stage).
The Pose Estimation stage registers the previously extracted model points to those of the live frame. This is commonly achieved by the Iterative Closest Point algorithm.
With regards to pertinent literature, I would advise the following papers as a starting point.
KinectFusion: Real-Time Dense Surface Mapping and Tracking
Real-time 3D Reconstruction at Scale using Voxel Hashing
Very High Frame Rate Volumetric Integration of Depth Images on
Mobile Devices
Related
I am looking at the Clubs Dataset https://clubs.github.io/ for some research into multiway registration of point clouds. I am initially trying to use the ICP registration method in a sequential order adding one point cloud at a time.
The dataset has the RGB parameters for the various poses of the object.
1525694104,-0.0913515162993169,-0.16815189159691318,0.4504956847817425,0.4591556084159897,-0.7817619820248951,-0.3682132612946675,0.20601769355692517
1525694157,-0.22510740390250225,-0.32514596548025265,0.45221561140129063,0.2388698281592328,-0.8750788198918591,-0.4081451880445991,0.10293936130543749
1525694174,-0.4179094161019803,-0.39403349319958664,0.4522321523188167,-0.004021371419719342,0.9070013543525104,0.4210736433584828,0.005455788205342825
The columns are namely
filename, translation across 3 axes, quaternion for rotation.
I converted the images into point clouds and when I try to align the point clouds, I would need to have a good estimation of the transformation. Given the measurements at two points, how would I find out the transformation between those points. I would use that as my intiial estimate of my ICP registration algorithm.
I need to segment depth image that captured from
a kinect device in realtime(30fps).
Currently I am using EuclideanClusterExtraction from PCL, it works but very slow(1fps).
Here is a paragraph in the PCL tutorial:
“Unorganized” point clouds are characterized by non-existing point references between points from different point clouds due to varying size, resolution, density and/or point ordering. In case of “organized” point clouds often based on a single 2D depth/disparity images with fixed width and height, a differential analysis of the corresponding 2D depth data might be faster.
So I think there are faster method to segment depth image.
The project doesn't use the RGB Camera, so I need a segmentation method that use only the depth image.
PCL provides segmentation algorithms optimised for organised point clouds.
For details see:
The tutorial here describing them and showing how to use them:
http://www.pointclouds.org/assets/icra2012/segmentation.pdf
The example code in thePCL distribution (relatively late versions): organized_segmentation_demo and openni_organized_multi_plane_segmentation
In the API, OrganizedConnectedComponentSegmentation and OrganizedMultiPlaneSegmentation. The latter builds on the former.
I have an accurate mesh surface model of an implant I'd like to optimally rigidly align to a computed tomography scan (scalar volume) that contains the exact same object. I've tried detecting edges in the image volume with canny filter and doing an iterative closest point alignment between the edges and the vertices of the mesh, but it's not working. I also tried voxelizing the mesh, and using image volume alignment methods (Mattes Mutual) which yields very inconsistent results.
Any other suggestions?
Thank you.
Generally, mesh and volume are two different data structures. You have to either convert mesh to volume or convert volume to mesh.
I would recommend doing a segmentation of volume data first, to segment out the issues you want to register. With canny filter might not be enough to segment the border clearly. I would like to recommend you with level-set method and active contour model. These two are frequently used in medical image processing. For these two topics, I would recommend professor Chunming Li's work.
And after you do the segmentation of volume data, you might be able to reconstruct mesh model of that volume with marching cubes. The vertexes of two mesh could be registered through a simple ICP algorithm.
However, this is just a workaround instead of real registration, it always takes too much time to do the segmentation.
I have a project where I have to recognize an entire room so I can calculate the distances between objects (like big ones eg. bed, table, etc.) and a person in that room. It is possible something like that using Microsoft Kinect?
Thank you!
Kinect provides you following
Depth Stream
Color Stream
Skeleton information
Its up to you how you use this data.
To answer your question - Official Micorosft Kinect SDK doesnt provides shape detection out of the box. But it does provide you skeleton data/face tracking with which you can detect distance of user from kinect.
Also with mapping color stream to depth stream you can detect how far a particular pixel is from kinect. In your implementation if you have unique characteristics of different objects like color,shape and size you can probably detect them and also detect the distance.
OpenCV is one of the library that i use for computer vision etc.
Again its up to you how you use this data.
Kinect camera provides depth and consequently 3D information (point cloud) about matte objects in the range 0.5-10 meters. With this information it is possible to segment out the floor (by fitting a plane) of the room and possibly walls and the ceiling. This step is important since these surfaces often connect separate objects making them a one big object.
The remaining parts of point cloud can be segmented by depth if they don't touch each other physically. Using color one can separate the objects even further. Note that we implicitly define an object as 3D dense and color consistent entity while other definitions are also possible.
As soon as you have your objects segmented you can measure the distances between your segments, analyse their shape, recognize artifacts or humans, etc. To the best of my knowledge however a Skeleton library can recognize humans after they moved for a few seconds. Below is a simple depth map that was broken on a few segments using depth but not color information.
I hope to find some hints where to start with a problem I am dealing with.
I am using a Kinect sensor to capture 3d point clouds. I created a 3d object detector which is already working.
Here my task:
Lets say I have a point cloud 1. I detected a object in cloud A and I know the centroid position of my object (x1,y1,z1). Now I move my sensor around a path and create new clouds (e.g. cloud 2). In that cloud 2 I see the same object but e.g. from the side, where the object detection is not working fine.
I would like to transform the detected object form cloud 1 to cloud 2, to get the centroid also in cloud 2. For me it sound like I need a matrix (Translation, Rotation) to transform point from 1 to 2.
And ideas how I could solve my problem?
Maybe ICP? Are there better solutions?
THX!
In general, this task is called registration. It relies on having a good estimation of which points in cloud 1 correspond to which clouds in point 2 (more specifically, which given a point in cloud 1, which point in cloud 2 represents the same location on the detected object). There's a good overview in the PCL library documentation
If you have such a correspondence, you're in luck and you can directly compute a rotation and translation as demonstrated here.
If not, you'll need to estimate that correspondence. ICP does that for approximately aligned point clouds, but if your point clouds are not already fairly well aligned, you may want to start by estimating "key points" (such as book corners, distinct colors, etc) in your point clouds, computing a rotation and translation as above, and then performing ICP. As D.J.Duff mentioned, ICP works better in practice on point clouds that are already approximately aligned because it estimates correspondences using one of two metrics, minimal point-to-point distance or minimal point to plane distance, according to wikipedia, the latter works better in practice, but it does involve estimating normals, which can be tricky. If the correspondences are far off, the transforms likely will be as well.
I think what you were asking about was in particular to the Kinect Sensor and the API Microsoft released for it.
If you are not planning to do reconstruction, you can look into the AlignPointClouds function in Sensor Fusion namespace. This should take care of it automatically, in methods similar to the answer given by #pnhgiol.
On the other hand, if you are looking at doing reconstruction as well as point cloud transforms, the Reconstruction class is what you are looking for. All of which can be found out about, here.