Technical data

Autostar IP Appendix 39
-39-
Appendix A
Image Processing Basics
Image processing allows you to modify the appearance of an image by applying various types of filters,
scaling options or transformations. The simplest type of processing is linear scaling. With linear scaling,
one pixel from the source image is multiplied by a scale factor, then an offset term is added. The original
pixel value is then replaced with the resulting value. This process is repeated on each pixel in the image.
The contrast and brightness of the image can be controlled by varying the scale factor (contrast) and the
offset value (brightness). You should attempt to set the contrast to a value that allows the full range of the
important parts of the image to fall on the sloping section of the transfer curve (see the Scale Image
function), then set the brightness so that the minimum pixel values appear black or very dark gray. The
following histogram and transfer graph illustrates the optimal contrast and brightness settings to give a
normal view of the image.
You may need to experiment to find the right settings for any given image. Sometimes it may be
necessary to scale different areas of the image using
Another useful image processing technique, convolution, allows you to enhance the appearance of an
image by 'filtering' the image data by either smoothing (low-pass filter) or sharpening (high-pass filter) or
combining parts of both types of filters.
This is accomplished by passing a convolution matrix, or kernel, over the entire image and replacing the
center pixel in the resulting matrix with the scaled sum of all of the other values. For example, if you
wanted to average each pixel with its adjacent neighbors the following kernel could be used.
11 1
1 1 1
11 1
Each pixel, including the center one, would be multiplied by 1, then the sum of each of these terms is
calculated. The result is then divided by the sum of each of the kernel values, in this case 9. Finally, the