unexpected behaviour of filter with multipe arguments?

[wiesehahn]

I'm using LASlib the library developed by LAStools to read and write LAS ans LAZ files. Sometime I do no know all the features myself. I'm not sure how these features are working and are poorly documented in LAStools itself. I don't know if filter_and is even useful because -keep_first -keep_class 2 is already an AND operation.

I just tested it and we can do -keep_first -keep_class 2 -filter_and, and also -keep_first -keep_class 2 -filter_or. Actually the equivalent of -keep_first -keep_class 2 is -keep_first -keep_class 2 -filter_or by which we get all points which are EITHER first returns OR ground points. By -keep_first -keep_class 2 -filter_and we get all points which are first returns AND classified as ground.

Originally posted by @wiesehahn in #46 (reply in thread)

I just came across the filter_and and filter_or options again.
While the statements

the equivalent of -keep_first -keep_class 2 is -keep_first -keep_class 2 -filter_or

seems true,

the statements that

-keep_first -keep_class 2 -filter_or by which we get all points which are EITHER first returns OR ground points

seems not true

With drop the behaviour fits my understanding, but with keep I think its weird. I am a little confuse now, maybe everything is correct, but maybe not. I cant clearly think about it anymore, but possibly the following examples give some hints on what I thought was going on when using multiple filters.
Some notes by Martin about the behaviour with drop (which seems correct anyway) here

library(lasR)
f <- system.file("extdata", "Megaplot.laz", package="lidR")


# intended: keep points which are first returns and also keep points which are of classification 2
# actual: keep points which are first returns and of classification 2
f1 <- exec(reader_las(filter = "-keep_first -keep_class 2") + write_las(), on = f) |> 
  lidR::readLAS()
summary(as.factor(f1@data$Classification))
#>    2 
#> 5032
summary(as.factor(f1@data$ReturnNumber))
#>    1 
#> 5032

# intended: keep points which are first returns and also keep points which are of classification 2
# actual: keep points which are first returns and of classification 2
f2 <- exec(reader_las(filter = "-keep_first -keep_class 2 -filter_or") + write_las(), on = f) |> 
  lidR::readLAS()
summary(as.factor(f2@data$Classification))
#>    2 
#> 5032
summary(as.factor(f2@data$ReturnNumber))
#>    1 
#> 5032

# intended: keep points which are first returns and of classification 2
# actual: keep points which are first returns and keep points which are classification 2
f3 <- exec(reader_las(filter = "-keep_first -keep_class 2 -filter_and") + write_las(), on = f) |> 
  lidR::readLAS()
summary(as.factor(f3@data$Classification))
#>     1     2 
#> 50724  7389
summary(as.factor(f3@data$ReturnNumber))
#>     1     2     3     4 
#> 55756  1281   929   147



# intended: drop points which are first returns and drop points which are of classification 2 (drop points which are either first returns or of classification 2)
# actual: same
f4 <- exec(reader_las(filter = "-drop_first -drop_class 2") + write_las(), on = f) |> 
  lidR::readLAS()
summary(as.factor(f4@data$Classification))
#>     1 
#> 23477
summary(as.factor(f4@data$ReturnNumber))
#>     2     3     4 
#> 20212  3070   195

# intended: drop points which are first returns and drop points which are of classification 2 (drop points which are either first returns or of classification 2)
# actual: same
f5 <- exec(reader_las(filter = "-drop_first -drop_class 2 -filter_or") + write_las(), on = f) |> 
  lidR::readLAS()
summary(as.factor(f5@data$Classification))
#>     1 
#> 23477
summary(as.factor(f5@data$ReturnNumber))
#>     2     3     4 
#> 20212  3070   195

# intended: drop points which are first returns AND also of classification 2
# actual: same
f6 <- exec(reader_las(filter = "-drop_first -drop_class 2 -filter_and") + write_las(), on = f) |> 
  lidR::readLAS()
summary(as.factor(f6@data$Classification))
#>     1     2 
#> 74201  2357
summary(as.factor(f6@data$ReturnNumber))
#>     1     2     3     4 
#> 50724 21493  3999   342

[wiesehahn]

Ok, I found some explanation (and also the confusion)
https://groups.google.com/g/lastools/c/YQ_suFQ-C9c/m/m8jmkUr2AQAJ

[Jean-Romain]

Thank you for sharing. Honestly I'm more confused now than before. The boolean logic seems inverted. For simple cases you should see the filters as a pipeline. In -keep_first -keep_class 2 the first returns pass the first test and are tested for class 2 while non first return are discarded and not even tested for ground. This is effectively an AND operation. However seems to be equivalent to -keep_first -keep_class 2 -filter_or. From that point this gives me a headache.

[wiesehahn]

This is exactly what I was struggling with. But at least now I know how it works, and because it is confusing I might remember :)