[R-sig-Geo] Surface Interpolation method for DEM point retrived by Terrestrial Laser Scanning after a outlier removing procedure

[gianni lavaredo]

Dear Researchers,

I wrote a code in R to processing around 1.2 GB of laser data derived by a
Terrestrial Laser Scanning in a forested area. There are few techniques to
derive a DEM from the ground points. One is create a grid of 0.5 x 0.5 m
and give for each pixel the minimum Z value of the points inside the voxel
(0.5 x 0.5 m by the Hmax of vegetation). Normally,  In this ground points
there are a component of outlier. The normal techniques proposed are:


[image: Inline image 1]



1. remove manually
2. special filter based on the angle of neighboured points
3 . RANSAC

 I used locoutNeighbor of library(mvoutlie) with good result using these
set parametrs:

Myres <-
locoutNeighbor(as.matrix(ppZmin.df.clean$z),ppZmin.df.clean$x,ppZmin.df.clean$y,
    variant="knn",
    usemax=1,
    chisqqu=0.999,
    propneighb=1,npoints=nrow(ppZmin.df.clean))


I just make some try yesterday night

[image: Inline image 2]

I am interest to comapre on my data more Outlier removing techniques and
any suggest is really appreciate

Thanks in advance
Gianni
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20120709/7029d728/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/png
Size: 12004 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20120709/7029d728/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/png
Size: 11998 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20120709/7029d728/attachment-0001.png>