trimImage: Trim zero rows or columns from an object of class 'Image'.

In Ecfun: Functions for 'Ecdat'

Trim zero rows or columns from an object of class Image.

Description

Identify rows or columns of a matrix or 3-dimensional array that are all 0 and remove them.

Usage

trimImage(x, max2trim=.Machine$double.eps, 
      na.rm=TRUE, returnIndices2Keep=FALSE, 
      ...)

Arguments

x	a numeric matrix or 3-dimensional array or an object with subscripting defined so it acts like such.
max2trim	a single number indicating the max absolute numeric value to trim.
na.rm	logical: If TRUE, NAs will be ignored in determining the max absolute value for the row. If a row or column is all NA, it will be treated as all 0 in deciding whether to trim. If FALSE, any row or column containing an NA will be retained.
returnIndices2Keep	if TRUE, return a list with 2 integer vectors giving row and column indices to use in selecting the desired subset of x. This allows an array y to be trimmed to match x. If FALSE, return the desired trimmed version of x. If this is a list with two two integer vectors, use them to trim x.
...	Optional arguments; not currently used.

Details

1. Check arguments: 2 <= length(dim(x)) <= 3? is.logical(na.rm)? returnIndices2Keep = logical or list of 2 integer vectors, all the same sign, not exceeding dim(x)?

2. if(is.list(returnIndices2Keep)) check that returnIndices2Keep is a list with 2 integer vectors, all the same sign, not exceeding dim(x). If yes, return x appropriately subsetted.

3. if(!is.logical(returnIndices2Keep)) throw an error message.

4. Compute indices2Keep.

5. If(returnIndices2Keep) return (indices2Keep) else return x appropriately subsetted.

Value

if(returnIndices2Keep==TRUE) return a list with 2 integer vectors to use as subscripts in trimming objects like x.

Otherwise, return an object like x appropriately trimmed.

Author(s)

Spencer Graves

See Also

trim trims raster images, similar to trimImage.

trimws trims leading and trailing spaces from character strings and factors. Similar trim functions exist in other packages but without obvious, explicit consideration of factors.

Examples

##
## 1.  trim a simple matrix
##
tst1 <- matrix(.Machine$double.eps, 3, 3,
    dimnames=list(letters[1:3], LETTERS[1:3]))
tst1[2,2] <- 1
tst1t <- trimImage(tst1)

# check
tst1. <- matrix(1, 1, 1,
          dimnames=list(letters[2], LETTERS[2]))

all.equal(tst1t, tst1.)


##
## 2.  returnIndices2Keep
##
tst2i <- trimImage(tst1, returnIndices2Keep=TRUE)
tst2a <- trimImage(tst1, returnIndices2Keep=tst2i)

tst2i. <- list(index1=2, index2=2)


# check

all.equal(tst2i, tst2i.)


all.equal(tst2a, tst1.)


##
## 3.  trim 0's only
##
tst3 <- array(0, dim=3:5)
tst3[2, 2:3, ] <- 0.5*.Machine$double.eps
tst3[3,,] <- 1

tst3t <- trimImage(tst3, 0)

# check
tst3t. <- tst3[2:3,, ]

# check

all.equal(tst3t, tst3t.)


##
## 4.  trim NAs
##
tst4 <- tst1
tst4[1,1] <- NA
tst4[3,] <- NA

tst4t <- trimImage(tst4)
# tst4o == tst4
tst4o <- trimImage(tst4, na.rm=FALSE)

# check

all.equal(tst4t, tst1[2, 2, drop=FALSE])



all.equal(tst4o, tst4)


##
## 5.  trim all
##
tst4a <- trimImage(tst1, 1)

tst4a. <- matrix(0,0,0,
     dimnames=list(NULL, NULL))


all.equal(tst4a, tst4a.)

Ecfun documentation built on Oct. 10, 2022, 1:06 a.m.