NAME

s.edgedetection

SYNOPSIS

s.edgedetection
s.edgedetection help
s.edgedetectioninput=string outClas=string Sge=float Sgn=float [lambda_g=float] [Tg=float] [tg=float] [theta_g=float] [lambda_r=float]

DESCRIPTION

s.edgedetection allows in a digital surface model the detection of the edges of the objects: an edge is a boundary between two different regions, i.e. a significant change in the height value corresponding to a small shift of the horizontal position.

The command is part of a procedure to filter airbone laser scanning (LIDAR) data in order to extract Digital Terrain Model.
This procedure involves the execution of the following commands:
1) s.edgedetection - 2) s.growing - 3) s.correction - 4) s.bspline.reg

OPTIONS

The user can run this program either interactively or non- interactively. The program is ran non-interactively if the user specifies program arguments and flag settings on the command line using the form:

s.edgedetection input=string outClas=string Sge=float Sgn=float [lambda_g=float] [Tg=float] [tg=float] [theta_g=float] [lambda_r=float]

Alternately, the user can simply type s.edgedetection on the command line without program arguments. In this case, the user is prompted for parameter values and flag settings using the standard GRASS parser interface described in the manual entry for parser.

Parameters:

input=string
input observation sites list file name
outClas=string
output classification sites list file name
Sge=float
interpolation spline step value in east direction
Sgn=float
interpolation spline step value in north direction
lambda_g=float
regularization weight in gradient evaluation
Default: 0.01
Tg=float
high gradient threshold for edge classification
Default: 6
tg=float
low gradient threshold for edge classification
Default: 3
theta_g=float
angle treshold to identify the direction
Default: 0.26
lambda_r=float
regularization weight in residual evaluation
Default:2

NOTES

The input in LIDAR data analisys is the last pulse site list file.

The edge detection procedure involves two different spline interpolations: a bilinear one and a bicubic one with different regularization weight values (lambda_g and lambda_r). [see s.bspline.reg]

The gradient magnitude (G), derived from the bilinear spline surface, the bias value (S), between the bicubic spline surface and the raw data, and the maximum slope direction (D) are computed for every sites of the input.

Knowing that:
- an edge is characterized by a hight gradient magnitude, a positive bias value and to belong to a chain element
- a point belongs to a chain element if its nearest points have similar D values along the direction perpendicular to the maximum slope direction

Taking into account a point P, being P1 and P2 the two nearest points of P and D1 and D2 their maximum slope direction, P is classified as object if:
(i) G>Tg and S>0
or
(ii) G>tg, S>0, (D-D1)<theta_g and (D-D2) <theta_g and there is almost one point in the nearest eight<of P with G>Tg

Every point in the input site file is examinated and reclassified as:
·       Terrain (0)
·       Object (1)
·       Uncertain (2)

input format example (x|y|z):

513629.21|5403205.11|#0 %291.24
513629.25|5403206.53|#0 %291.28
513629.29|5403208.16|#0 %291.27
513629.33|5403209.69|#0 %291.29
513629.37|5403211.12|#0 %291.32

outClas format example (x|y|z%classification%bicubic_interpolated_value):
 
499827.48|5418882.79|#0 %247.49 %0 %247.06513365
499827.1|5418870.68|#0 %246.7 %0 %246.32005415
499826.72|5418858.43|#0 %245.7 %1 %245.59059607
499826.33|5418846.07|#0 %244.65 %0 %244.9670436
499825.96|5418833.6|#0 %244.23 %2 %244.52604082

SEE ALSO

s.correction, s.bspline, s.growing, s.to.qrast

AUTHORS

Maria A. Brovelli, Politecnico di Milano - Campus Como
Massimiliano Cannata, Politecnico di Milano - Campus Como
Ulisse M. Longoni, Politecnico di Milano - Campus Como

REFERENCES

Brovelli M.A.; Cannata M.; Longoni U.M., LIDAR Data Filtering and DTM Interpolation Within GRASS
 Transactions in GIS, April 2004, vol. 8, no. 2, pp. 155-174(20)  - Blackwell Publishing Ltd

Pratt K 1991 Digital image processing (Second edition). New York, NY, John Wiley and Sons