Live 3D Point Cloud Stitching Demo Video for Kinect with MATLAB

Today I'll share the Kinect Live stitching demo Program as below.

By using GUI Program which I introduced in last post, we can acquire RGB stream, Depth stream and real-time 3D reconstructed stream.

Also, by moving the device, we can build the real-time 3D reconstruction stream, as following demo. In other words, 3D point cloud stitching process for Kinect data can be done in real-time.
Note that, 3D stitching time(Number of frames) is controllable.

Thanks for reading.
Will come back with more interesting topic next time:)

FYI, you can get above GUI program from below.
Download Live 3D Point cloud for Kinect Demo

Wanbin Song

Building Handy GUI program for Kinect with MATLAB

With simple GUI building skills, you can easily make handy program with MATLAB(R2015b). You can do :

• Visualize streaming Color and Depth image.

• Control depth mode(Default/Near) and elevation angle.

• Simultaneously align color with depth image.

• Live 3D Reconstruction with Point cloud stitching

The highlight thing in this demo is live 3D reconstruction. User can easily combine multiple point clouds to reconstruct a 3D scene using ICP(Iterative Closest Point) algorithm. It can be used in various application like develop 3D models of objects or build 3D world maps for SLAM(Simultaneous Localization And Mapping).

This is all for my recent work on Kinect. Any opinions to discuss are welcome.

Thanks for reading.

Wanbin Song

Kinect v2 Preview(3)

After a long absence, today, i'll discuss about imperfections of Kinect v2.

As i mentioned in last posting, due to the change in method of measuring depth, there exists several differences between Kinect v1's depth image with that of Kinect v2.

Usually, there were large amounts of holes in Kinect v1's depth image, including random holes and occlusions.

Left image of above figure shows depth image of Kinect v1. As you can see, occlusions and random holes(white) are very thick. Comparing to this image, right one is depth image of new version, which captures the same scene with Kinect v1's. And we can figure out that holes are mainly exists along the object boundaries, with the small amount. This is the first major difference between two depth image.

There is another noticeable difference between two. As in above right image, on the corner, there exists large blobs of random holes which disappears in Kinect v1's depth image. This is because intensity values of of returned infrared light are week. But these random holes exists corner area of depth image, not the region of interest.

So, Kinect v2's hole-filling problem is not an important issue, because the quality of depth image is improved significantly.

This is the conclusion of my recent work on Kinect v2's depth image. Any opinions of yours are welcome, and let us discuss about it.

Thanks,

------------------------------------------------------------------

Song, Wanbin

M.S. Candidate

Laboratory of Digital Image Processing

Department of Electronic and Electrical Engineering

Dongguk Univ. Seoul, Korea

Homepage   : http://dip.dongguk.edu

E-mail        : wbsong@dongguk.edu
                   me.wbsong@gmail.com

------------------------------------------------------------------

Kinect v2 Preview(2)

Today, I will discuss about the basic properties of Kinect version 2.

Distance measuring method is changed from structured light pattern(Kinect v1) to
ToF(Time of Flight)(Kinect v2) as mentioned in previous post, several changes
can be detected in depth image of new Kinect.

First of all, there is no interference between multiple Kinects.
Previous version of Kinect use structured light pattern for depth measuring, so
if another kinect device is running simultaneously, holes are detected in overlapping area as below figure. White blobs in below depth image shows interference which has unknown depth value, and it is called as hole.

<Holes in depth image>

But, there is no interference between multiple devices using Kinect v2.
First image is the depth image when only single kinect is processing, and second one is the depth image when multiple devices are running.

<Single Sensor>

<Multiple Sensor>

You can see there is no difference between two depth image, so we can use multiple kinect w/o interference in Kinect v2.

You can also see the edge of depth image remains black, and I think that phenomena is the property of tof.

Another big change in Kinect v2 is infrared property.

In old Kinect, Kinect IR camera can sense out door and cannot measures depth in outdoor. But, using Kinect v2, it can operate in outdoor.

As above figure, with the ir sensor of each Kinect, we can see outdoor with Kinect v1 but cannot with Kinect v2. I think Kinect v2 uses there own frequency, so it cannot sense outdoor but with there own frequency, it can measure the depth in outdoor as below figure.

It is quite important change in new Kinect, and with longer working range in Kinect v2(up to 8m), various applications can be developed in outside.

This is all for second preview of Kinect version 2.

Next post will coming soon with more informative data.

Thank you for reading.

------------------------------------------------------------------

Song, Wanbin

M.S. Candidate

Laboratory of Digital Image Processing

Department of Electronic and Electrical Engineering

Dongguk Univ. Seoul, Korea

Homepage   : http://dip.dongguk.edu

E-mail        : wbsong@dongguk.edu

                   me.wbsong@gmail.com

------------------------------------------------------------------

Kinect v2 Preview(1)

Microsoft

The New Kinect is released.

In this blog, I'll introduce some information about New Kinect
which is Kinect v2, and share some codes with anyone who is interested in Computer vision.

The performance of Kinect v2 is improved remarkably comparing to the previous version.

First of all, measuring depth method is changed from structured light pattern to TOF(Time of Flight).
Therefore, the density of depth image is quite greater than previous one and the resolution of color image is also improved to FHD.

The detail specification is as following.

Resolution (pixel)

	Kinect v1	Kinect v2
Color	640x480	1920x1080
Depth	320x240	512x424

FOV (deg)

	Kinect v1	Kinect v2
Color	62x48.6	84.1x53.8
Depth	57.5x43.5	70.6x60

And the remarkable improvement is the number of people of skeletal tracking simultaneously.
In Kinect v1, only two people can tracked at the same time, but by Kinect v2, six people can be tracked.
And they additionally provide thumb tacking or end of hand tracking.
I'll discuss detail in few days.
Lastly, the working range is improved due to the method of measuring depth.
Using TOF method, the working range in Kinect v2 is 0.5m ~ 8m.

Up to here, I briefly mentioned about characteristics of Kinect v2.

From the next post, let's share some codes and practice some applications using Kinect SDK v2.

Thank you for reading.

------------------------------------------------------------------

Song, Wanbin

M.S. Candidate

Laboratory of Digital Image Processing

Department of Electronic and Electrical Engineering

Dongguk Univ. Seoul, Korea

Homepage   : http://dip.dongguk.edu

E-mail        : wbsong@dongguk.edu

                   me.wbsong@gmail.com

------------------------------------------------------------------