chunk: Get indices for partial image reading/writing

In nplatonov/ursa: Non-Interactive Spatial Tools for Raster Processing and Visualization

Get indices for partial image reading/writing

Description

In the case of 'Cannot allocate vector of size ...' error message chunk_band returns list of bands indices, which are suitable for allocation in memory at once, chunk_line returns list of lines (rows) indices, which are suitable for allocation in memory at once. chunk_expand is used to expand lines indices and can by applied in focal functions.

Usage

chunk_band(obj, mem = 100, mul = 1)
chunk_line(obj, mem = 100, mul = 1)
chunk_expand(ind, size = 3)

Arguments

obj	Object of class ursaRaster
mem	Numeric. Memory size in GB, which is suitable for allocation.
mul	Numeric. Expansion or reduction factor (multiplier) of default value of memory allocation.
ind	Integer. Line indices.
size	Integer. Size of focal window.

Value

chunk_band returns list with sequences of bands

chunk_line returns list with sequences of lines

chunk_expand returns list:

src	expanded set if line indices
dst	matching of source indices in the expanded set

Author(s)

Nikita Platonov [e-mail]

Examples

 ## 1. Prepare data
 session_grid(NULL)
 fname <- ursa:::.maketmp(2)
 a <- create_envi(fname[1],nband=3,ignorevalue=-99)
 for (i in seq(nband(a)))
   a[i] <- pixelsize()^(1/i)
 close(a)
 rm(a)
 
 ## 2. Read 
 a <- open_envi(fname[1])
 chB <- chunk_band(a,2)
 str(chB)
 for (i in chB)
    print(a[i])
 chL <- chunk_line(a,2.5)
 str(chL)
 for (j in chL)
    print(a[,j])
 
 ## 3. Filtering with partial reading
 b <- create_envi(a,fname[2])
 fsize <- 15
 for (j in chL) {
    k <- chunk_expand(j,fsize)
    b[,j] <- focal_mean(a[,k$src],size=fsize)[,k$dst]
 }
 d1 <- b[]
 
 ##  4. Filtering in memory
 d2 <- focal_mean(a[],size=fsize)
 close(a,b)
 envi_remove(fname)
 print(d1-d2)
 print(round(d1-d2,4))

nplatonov/ursa documentation built on May 8, 2024, 6:02 p.m.