[ros-users] Hokuyo intensity values

Radu Bogdan Rusu rusu at willowgarage.com
Wed Apr 14 15:34:02 UTC 2010


I concur with what Vijay said. The intensity data was used for extracting the door handle using statistical methods.

Regarding intensity calibration, there's work mostly done by our colleagues in photogrammetry and sensing 
(http://www.gc.vgtu.lt/upload/geod_zurn/gc_vol35_no3_77-81_vain.pdf for example).

Cheers,
Radu.

Vijay Pradeep wrote:
> Hi Abe,
> 
> It's great to hear that you're interested in using Hokuyo intensity 
> data.  I'm not certain as to how much intensity data is being used in 
> our point cloud processing pipelines.  I think intensity might be used 
> in our door handle detector.  Someone else can elaborate further on this.
> 
> Intensity data is a critical part of the pr2_calibration stack.  We use 
> intensity images (just like what was on the 
> 'Scanner_to_camera_calibration' page) to extract checkerboard corners, 
> which is the first step in a much larger full system calibration procedure.
> 
> /it seems that the values are normally somewhere between 0 and 6000. 
> However every once in a while, they'll go as high as ~15000./
> I can agree with the 0-6000 intensity estimate, but I've definitely also 
> seen spurious bright returns.  I haven't really paid attention to how 
> large these returns get (maybe they are in fact ~15000), but I have 
> noticed that the spurious bright returns occur most often when rays 
> glance off the edges of objects.
> 
> /it says that the intensity is a function of distance... is there a good 
> way to normalise out the distance factor so that intensity values can be 
> compared on a normalised 0 to 1 range?
> /I don't believe anyone here has tried to build an intensity/range 
> model, probably because we don't have a strong use case for this type of 
> normalization.  Somewhat related, I'd be curious to see how intensity 
> data could be used to compensate for range biases.
> 
> I'm definitely interested in hearing how you end up using the intensity 
> data!
> 
> Vijay
> 
> On Tue, Apr 13, 2010 at 4:57 PM, Abe Bachrach <abachrach at gmail.com 
> <mailto:abachrach at gmail.com>> wrote:
> 
>     Hi there,
>     I started looking at the intensity values that you can get from the
>     hokuyo UTM laser scanners, but was seeing a much wider fluctuation
>     in the values than I would have expected...
>     from looking at my own data, and the data posted on the calibration wiki
>     http://pr.willowgarage.com/wiki/Scanner_to_camera_calibration
> 
>     it seems that the values are normally somewhere between 0 and 6000.
>     However every once in a while, they'll go as high as ~15000.
> 
>     In the documentation posted here:
>     http://www.ros.org/wiki/hokuyo_node?action=AttachFile&do=get&target=UTM-30LX_Specification.pdf
>     <http://www.ros.org/wiki/hokuyo_node?action=AttachFile&do=get&target=UTM-30LX_Specification.pdf>
>     it says that the intensity is a function of distance... is there a
>     good way to normalise out the distance factor so that intensity
>     values can be compared on a normalised 0 to 1 range?
> 
>     How have you guys generally worked with the intensity info.
> 
>     thanks in advance!
>     -=Abe
> 
> 
>     _______________________________________________
>     ros-users mailing list
>     ros-users at code.ros.org <mailto:ros-users at code.ros.org>
>     https://code.ros.org/mailman/listinfo/ros-users
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users

-- 
| Radu Bogdan Rusu | http://rbrusu.com/



More information about the ros-users mailing list