classSegVal: Validate a Tree Prediction with TreeCrown Segments

In SchoenbergA/IKARUS: Classification Tool

Description

calculates the amount of cells for class "trees" which overlap with TreeCrown Segments.

Usage

classSegVal(pred, seg, classTree = NULL, reclass = NULL)

Arguments

pred	RasterLayer - Prediction to be validated
seg	PolygonLayer - Segments with TreeCrowns
classTree	numeric - The class representing Trees
reclass	optional - numeric value of class to merge into class "trees", for multiple classes to merge use reclass=c(x,y)

Value

returns the validation score

• return and plot the "tree 1 and no tree 2" layer, will be reclassified if 'reclass' is used.

• result table

• nclass - amount of cells for class 'tree' (may differ if 'reclass' is used)

• nseg - amount of cells for all TreeCrowns

• overclass - amount of cells for classification not occuring in TreeCrowns (over classification)

• underclass - amount of cells for TreeCrowns not occuring in the classification (under classification)

• hit - amount of cells for classification occuring in TreeCrowns

• hitrate - amount of hit in relation to total classification cells in %

• rate underclass - amount of underclass in relation to total TreeCrown cells in %

• rate overclass -amount of overclass in relation to total classification cells in %

• validation score hitrate @ overrate + underrate

Author(s)

Andreas Schönberg

Examples

# load data
require(caret)
require(CAST)
require(doParallel)
require(raster)
require(IKARUS)
require(mapview)
lau_Stk <- raster::stack(system.file("extdata","lau_RGB.grd",package = "IKARUS"))
lau_tP <-rgdal::readOGR(system.file("extdata","lau_TrainPolygon.shp",package = "IKARUS"))
# handle CRS string
crs(lau_tP) <- crs(lau_Stk)
# load segments
lau_seg <-rgdal::readOGR(system.file("extdata","lau_TreeSeg.shp",package = "IKARUS"))
# handle CRS string
crs(lau_seg) <- crs(lau_Stk)
### extract values using 'exrct_Tdat' to generate training dataset
tDat <- exrct_Traindat(lau_tP,lau_Stk,"class")
# check for class column and predictor columns in input training dataset
head(tDat)
# classification
model1 <- RFclass(tDat = tDat,predCol = "default",predStk = lau_Stk,classCol = "class")
#check model and seg
plot(model1$prediction)
plot(lau_seg)

# test several combinations of classes for tress
tree <- classSegVal(  pred=model1$prediction,  seg=lau_seg,  classTree=4,  reclass=NULL)
tnsha <- classSegVal(  pred=model1$prediction,  seg=lau_seg,  classTree=4,  reclass=2)
tnshangras <- classSegVal(  pred=model1$prediction,  seg=lau_seg,  classTree=4,  reclass=c(2,3))

SchoenbergA/IKARUS documentation built on Sept. 8, 2021, 11:11 a.m.