Biomedical Engineering Reference
In-Depth Information
many other image parameters. The image name may also contain useful
information required in the output, which can be easily captured by
ImageJ using the 'getTitle()' command. Images come in various bit depths
from one bit binary to 24 bit colour. ImageJ can handle various bit depths
but errors will occur if these images are processed by functions that
require images at certain bit depths so it is important to appreciate this
requirement. For example, the 'Analyse particles' function can only work
on binary images in which objects have been segmented. As well as bit
depth, the images can be of any physical pixel dimensions providing
enough RAM has been allocated to ImageJ, which can be set in the menu
Edit/Options/Memory and Threads. When running ImageJ under a 64 bit
operating system, more RAM can be allocated than is possible with a 32
bit system. This is useful when working with large images or arrays of
data. When opening large image sequences, the problem of running out
of RAM memory can be averted by opening them in a virtual sequence.
An alternative way of handling large sets of images is to handle images
one at a time within a macro using the 'openNext' command.
The colour space of an image describes the gamut of colours and
because human vision is based on trichromatic perception, most of the
colour models use three values, others use more [9]. The colour models
all form a 3D space where the colour components describe a location (i.e.
colour) in that space. Images are opened as 24 bit images with 8 bits per
red, green, blue (RGB) channel. In other words, each channel is assigned
8 bits per pixel, which is 256 possible shades of that colour channel. A
colour space has to be described by a minimum of three different
parameters and many different models exist. Within ImageJ it is possible
to split an image up by either the RGB or hue, saturation, brightness
(HSB) channels into a stack of three images.
The questions as to why you need to be able to work in different colour
spaces and when is one more useful than another are often asked. The
need to be able to do this becomes evident in particular imaging tasks
where a particular parameter becomes critical to Segment your image.
For example, when trying to separate yellow objects from purple objects
using the RGB colour space this may result in a blurred segmentation;
however, by switching to HSB and using the hue channel then objects of
different colours can easily be segmented using this channel independently
of their brightness or saturation. Besides 24 bit colour, ImageJ can also
handle images with larger bit depths per pixel, for example 32 bit
greyscale images. Look up tables (LUTs) can also be applied to 8 bit
images to quickly make false colour images and ImageJ provides many
LUT pallets of 256 colours ready to apply to 8 bit greyscale images.
Search WWH ::
Custom Search