Image Processing Reference
In-Depth Information
Fig. 2.18 The relationship
between the intensity values
and the different shades of
gray
Fig. 2.19 Definition of the
image coordinate system
A gray-scale image (as opposed to a color image, which is the topic of Chap. 3)
is a 2D array of pixels (corresponding to the 2D array of cells in Fig. 2.13 ) each
having a number between 0 and 255. In this text the coordinate system of the image
is defined as illustrated in Fig. 2.19 and the image is represented as f(x,y) , where
x is the horizontal position of the pixel and y the vertical position. For the small
image in Fig. 2.19 , f( 0 , 0 )
19.
So whenever you see a gray-scale image you must remember that what you are
actually seeing is a 2D array of numbers as illustrated in Fig. 2.20 .
=
10, f( 3 , 1 )
=
95 and f( 2 , 3 )
=
2.4.1 The Region of Interest (ROI)
As digital cameras are sold in larger and larger numbers the development within
sensor technology has resulted in many new products including larger and larger
numbers of pixels within one sensor. This is normally defined as the size of the
image that can be captured by a sensor, i.e., the number of pixels in the vertical
direction multiplied by the number of pixels in the horizontal direction. Having a
large number of pixels can result in high quality images and has made, for example,
digital zoom a reality.
When it comes to image processing, a larger image size is not always a benefit.
Unless you are interested in tiny details or require very accurate measurements in
the image, you are better off using a smaller sized image. The reason being that
when we start to process images we have to process each pixel, i.e., perform some
math on each pixel. And, due to the large number of pixels, that quickly adds up
to quite a large number of mathematical operations, which in turn means a high
computational load on your computer.
Say you have an image which is 500
×
500 pixels. That means that you have
500
250 , 000 pixels. Now say that you are processing video with 50 images
per second. That means that you have to process 50
·
500
=
12 , 500 , 000 pixels
per second. Say that your algorithm requires 10 mathematical operations per pixel,
then in total your computer has to do 10
·
250 , 000
=
·
12 , 500 , 000
=
125 , 000 , 000 operations
Search WWH ::




Custom Search