Image-class {EBImage}R Documentation

Image class defintion and accessor methods

Description

Image is the base class to hold data and parameters of grayscale and RGB images in R. The class is derived from array and inherits all of its properties and methods, e.g. mathematical operations on arrays, subsetting, histograms etc. Image processing routines are defined additionally as S4 methods for objects of this class. In many cases they also return objects of class Image.

Constants

The following constant are defined in the package and should be used everywhere where image color mode needs to be specified.

    Grayscale = as.integer(0)
    TrueColor = as.integer(1)

Accessor methods

# S4 methods for class 'Image':

colorMode(x), colorMode(x) <- value
Gets and sets image color mode. value is one of the constants above. Default: Grayscale.
compression(x), compression(x) <- value
Gets and sets compression algorithm used when image saved in a format supporting compression. For exapmple, TIFF images support ZIP and LZW lossless compressions, whereas JPEG uses JPEG compression by default. Possible values of type character are 'NONE', 'LZW', 'ZIP', 'JPEG', 'BZIP' and 'GROUP4'. Some of these may be unavailable on your system if ImageMagick was compiles without their support. Default: 'LZW'.
features(x)
Is a read only method that returns a list or matrices with descriptors of detected objects. If object detection was executed on the image, an empty list is returned. Otherwise, the number of elements matches the number of frames in the image. Default: list().
fileName(x), fileName(x) <- value
Gets and sets the file name. When reading images this value is updated, when writing it can be used as default if no alternative is specified. When reading a stack, the first file name in the list of stack images will be assigned. Default: 'no-name'.
imageData(x), imageData(x) <- value
Gets and sets image data. Image data is a 3D array with the last dimension matching the number of 2D images in a stack. The atomic type of the array is defined by colorMode().
resolution(x), resolution(x) <- value
Gets and sets values of image resoultion: a numeric vector of two values of resolution in x and y dimensions. Generally resolution is understood in pixel-per-inch (ppi) or dots-per-inch (dpi), however some microscopes can save these values in images in different units. It is the responsibility of the user to identify the unit and keep the units comparable between different images in the same project if required: the package does not take care of this. The default values are equal for both directions, 2.5e+6 dpi.
sampleFilter(x), sampleFilter(x) <- value
Gets and sets the filtering algorithm used when resizing images using resize and other image transformation routines. Possibly this value will be removed and supplied as argument to the corresponding function. At the moment, possible values include: 'point', 'box', 'triangle', 'hermite', 'hanning', 'hamming', 'blackman', 'gaussian', 'quadratic', 'cubic', 'catrom', 'mitchell', 'lanczos', 'bessel' and 'sinc'. Default: 'lanczos'.

R does not have a concept of private/public class members, therefore, the following cannot be enforced and are just recommendations: use only accessor methods to get/set slot data; and consequently – everything that does not have an accessor method should not be accessed directly by the user. For this reason, we do not describe slots here: consider all slots as private. Package developers can have a look at the code and see the full object structure.

Creating objects

Objects can be created using new("Image") supplying slot data if necessary. Additionally, wrapper functions are defined to simplify the construction of Image objects in different situations.

See Image constructor and read.image function on details how to create images or read images from disk or network.

Details

Image data are stored as 3D arrays, with first two dimensions specifying image/frame size and third – number of frames. A single 2D image will have the number of frames 1. In subsetting, the value of drop parameter is set to FALSE, thus image dimensionality is preserved at 3.

At the moment, images can store data either as grayscale values or true color values. Grayscale values are stored in numeric or double arrays, whereas true color data are stored as integer arrays. I.e. the type of image array with grayscale data can be tested as: is.array(x) && is.numeric(x) && !is.integer(x) whereas for true color data the test would look like: is.array(x) && is.integer(x). Correspondingly, the data slots are created with array(as.numeric(val),...) for grayscales and array(as.integer(val),...) for true color images. Grayscale values are assumed to be in the range [0,1], although this is not a requirement. However, many image processing functions will also assume data in this range or will generate wrong results if the data do not fit the range.

Author(s)

Copyright (c) 2005-2006 Oleg Sklyar : osklyar@ebi.ac.uk, LGPL >= 2.1

See Also

Image, image.Image, display, channel

Examples

  if ( interactive() ) {
    ## Not run: creates a new Grayscale image of shifted stripes
    a <- Image( (0:100)*0.01, c(60,60) )
    display( a )
    print( a )

    ## Not run: converts image mode to TrueColor, compare print's
    b <- a
    colorMode( b ) <- TrueColor
    print( b )

    ## Not run: fills values with brightness > 50% with red
    b[ a > 0.5 ] = as.integer( 255 )
    display( b )
  }

[Package EBImage version 2.0.1 Index]