imWo

Two functions for the computation of perimetric visual complexity (Pelli, Burns, Farell, & Moore-Page, 2006) of letters, words, symbols etc.

CreateWordImages() creates png files of any kind of characters provided in a data frame. PerimetricComplexity() computes complexity values for any binary png image stored in a given folder. The optional argument 'add.pi' provides the option to correct for perimeter underestimation as proposed by Watson (2002).

Watson, A. (2012). Perimetric Complexity of Binary Digital Images. The Mathematica Journal, 14. https://doi.org/10.3888/tmj.14-5

Pelli, D. G., Burns, C. W., Farell, B., & Moore-Page, D. C. (2006). Feature detection and letter identification. Vision Research, 46(28), 4646–4674. https://doi.org/10.1016/j.visres.2006.04.023

---

Preparations

Download the zip file from github, unzip, define working directory, and run the code below.

source("create_word_images.R") 
source("perimetric_complexity.R")

library(png)
library(readr)
library(readxl)

# Under Windows font problems were solved by defining font
# family explicitly beforehand 
windowsFonts(Arial=windowsFont("TT Arial"))
windowsFonts(Times=windowsFont("TT Times New Roman"))
windowsFonts(Courier= windowsFont("TT Courier New"))

Example import Hindi words from text file

hindi <- data.frame(read_csv("Hindi_tv.txt", col_names = FALSE))
# the encoding of file names will most proably cause trouble, thats why...
hindi$item.label <- 1:nrow(hindi) # for optional file name in case of encoding issues

Example import German letters from text file

german <- data.frame(read_csv("alphabet_german.txt", col_names = FALSE))
german$item.label <- 1:nrow(german) # for optional file name in case of encoding issues

Example import Hebrew letters (works best with Excel files)

hebrew <- read_excel("hebrew_alephbet.xlsx")

---

Creates images of words

The function CreateWordImages() takes several - rather self-explanatory arguments for image appearance. Important: provide a name of the data frame from which letters or words should be taken (items.df) and written into png-files. Not all operating systems allow UTF-8 file names, a workaround is to specify another column in your items.df that contains image.file names.

CreateWordImages(items.df = hebrew, # provide data frame name; see import options above
                 image.width = 5,
                 image.height = 1,
                 image.resolution = 150,
                 print.color = "black",
                 background.color = "white",
                 font.type = "Arial",
                 item.column = 1, # provide column index
                 file.name.extension = ".png",
                 file.name.column = 2,
                 file.name.encoding = "native")

Computes perimetric complexity

PerimetricComplexity(image.path = "itemImages/",
                     add.pi = TRUE, # perimetric complexity according to Pelli et al. (2006) or divided by 4π
                     output.file.name = "hebrew_complexity.Rda")