Lesson 3. Correction and Binarization
of Color Images
Brief
description: During this lesson you will learn how to correct
color images by histogram, binarize a color image and correct binarization
results.
For detailed information on these subjects see the following sections
of the Quick Start guide:
Correction by histogram,
Binarization color and greyscale images, Filtering
Monochrome Images.
Correcting a color image by histogram
Using correction by histogram you can tune the image
brightness, hue and contrast. You can apply automatic correction by histogram
or tune the parameters manually.
Choose Open from the File
menu. Open the utrecht.dwg file, created during
the previous lesson from the Lesson_2-3 folder
- or -
Open a new AutoCAD document - with the Open command from
the rFile menu open the Utrecht.tif document
from Lesson_2-3 folder.
- Choose Equalize from the Image menu.
During this exercise we will use manual correction by
histogram. To correct an image manually you need to specify three values:
lightest and darkest levels and image gamma (which defines the position
of the middle brightness value relative to the current threshold values)
- for each channel or for Master channel.
- In the Equalize dialog box choose Master
from the Channel list.
- The lightest level is set using the light eyedropper
 ,
with which you can specify the colours to turn white. (click on one
of the darkest pixels in the white part of the image).
All pixels that are lighter that the specified value, will turn
white.
- The darkest level is set using the dark eyedropper
 ,
with which you can specify the colours to turn black. (click on one
of the lightest pixels in the black letters).
All pixels that are darker than the specified value will turn black.
- Move the middle marker of the Histogram (setting the image gamma)
to obtain the best result in the preview window.
Moving the middle slider to the left increases the gamma value; the
image becomes lighter.
Moving the middle slider to the right reduces the gamma value; the image
becomes darker.
- When you achieve a result you are happy with in the Preview window,
press the Apply button.
Binarizing a color image
Binarization allows you to extract pixels of specified
colors (or colors close to them) to a monochrome layer. To do this, you
need to specify colors on the original image to transfer to another layer.
During this exercise we use binarization to obtain a
monochrome layer of roads, which are yellow on this image.
- Choose Binarization from the rImage menu.
- Choose the binarization method - Range; Open
the Range tab in the Binarization dialog.
- Chose a range method - by HSV and channel
- Hue in the corresponding lists of the Range
tab.
- Specify a Layer and its color on which the new monochrome
image will be placed.
- Using one of these eyedroppers
specify a range of colors that form the roads; these colors are added
to the Colors list. The estimated binarization results
are shown in the preview window.
INFORMATION:
The first eyedropper is used for specifying a color for a specified
pixel, the second eyedropper is used for specifying the averaged color
of all pixels adjacent to the specified by you; the third eyedropper is
used for specifying the averaged color of pixels contained within a polygon
defined by you.
- You can delete incorrectly selected colors by pressing the Delete
button.
- Tune the results using the sliders located in Ranges.
- You can delete small speckles that appear using the Noise Reduction
slider - move it towards the right, and observe the changes in the Preview
window.
- When you achieve satisfactory results in the Preview window, press
Apply.
- You can save the binarization template for future use. To do this,
press the Template button, choose Save,
and in the Save template File dialog specify a name and
location, and then press the Save button.
Improving the monochrome layer obtained after binarization
To improve the monochrome layer obtained after binarization,
you can apply some of the WiseImage tools for enhancing monochrome images.
Let's apply the Hole Remover and Smoothing
filters to correct text areas.
To make this operation easier, turn off the visibility
of the AutoCAD layer that contains the color image.
Hole Remover
- Choose Hole Remover from the Filters
menu.
- Clear the Auto estimating checkbox, then press
the Measure Max Size button and specify the largest
hole on the image.
- Press Apply.
|
 |
Smoothing
- Choose Smoothing from the Filters menu.
- Using the sliders - Medianning & Threshold
- achieve the best result in the Preview window.
- Press OK.
|
|
|
|
Before applying Smoothing
|
After applying Smoothing
|
Thinning
If needed, you can apply the Thinning
filter to make the obtained roads thinner.
- Choose Thinning from the Filters
menu.
- Specify Number of Passes equal to '1'.
- Press OK.
|
 |
|
