I have spent some time on the image (300 DPI, JPEG) attached to the first
post. Look at the result.
You may try to get it by yourself as the project is attached.
The quality of source lines is shown well in the second post.
First of all, I created a new project based on the image 3.jpg. Coordinate
units are pixels, coordinate origin – in the lower left corner of the image
(I do not go into matters of correction, transformation, georeferencing, etc.).
Then I opened 3.jpg, saved it as 3_relief.tif and added to the project.
After that, 3_relief.tif was treated with a sequence of raster editing operations.
The aim was – to blur and delete sepia points and to enhance contrast
and homogeneity of relief contours.
Note – red and green tints predominate in lines of relief whereas blue
is minor.
It was important to have it in mind at histogram-based mask selection
for Binarization.
The sequence of operations before Binarization comprised
(see 3_relief.LOG in the project archive):
------------------ 3_relief.tif ------------------
1. Image 3_relief.tif changed:
Diffuse :
Factor = 50
Radius = 1
2. Image 3_relief.tif changed:
Unsharp Mask :
Factor = 200
Radius = 4
Threshold = 0
3. Image 3_relief.tif changed:
Diffuse :
Factor = 50
Radius = 1
4. Image 3_relief.tif changed:
Unsharp Mask :
Factor = 200
Radius = 4
Threshold = 0
5. Image 3_relief.tif changed:
Diffuse :
Factor = 100
Radius = 2
6. Binarization
B&W image extraction, 3 masks
------------------ End Log ------------------
Processed image before Binarization:
Вложение:
Комментарий к файлу: Растр перед бинаризацией
Start.jpg [ 56.72 Кб | Просмотров: 14091 ]
Binarization.
!!! SWITCH OFF the Combine Masks option!!!
Masks were formed for:
1. The main group of tints
2. Intersections of green and brown lines
3. Light areas (this mask had been done roughly, as I realized later)
Range selection for the main mask
Always remember that overlapping of all three color intervals
causes capture of grey tints, i.e. irrelevant pixels (rubbish).
Вложение:
Комментарий к файлу: Маска основного набора оттенков коричневого
Base_mask.jpg [ 61.12 Кб | Просмотров: 14095 ]
Mask for gap filling at crossings with green lines
Вложение:
Комментарий к файлу: Маска комбинационного цвета в местах пересечений с зелёными линиями
Green_mask.jpg [ 63.19 Кб | Просмотров: 14095 ]
Mask for light parts of lines
Вложение:
Комментарий к файлу: Маска для светлых участков линий (подобрана плохо - готовил грабы экрана по одному участку и поленился посмотреть на светлые области. В итоге, на идеальных коричневых линиях на белом фоне возникли разрывы.)
Light_mask.jpg [ 60.68 Кб | Просмотров: 14094 ]
Then I clicked the Binarize button to extract a black-and-white
image corresponding to these three masks, saved it as 3_relief_BW.tif,
and added to the project.
The following sequence of operations smoothed 3_relief_BW.tif
and helped to delete rubbish:
1. Color mode transformation into 24bpp
3. Image changed:
Diffuse :
Factor = 100
Radius = 2
4. Image changed:
Unsharp Mask :
Factor = 500
Radius = 7
Threshold = 49
5. Image changed:
Contras Enhancement :
Channel: RGB
Input values 219/223
Output values 0/1.00/255
6. Color mode transformation into B&W (1bpp)
At the fourth step, the image looks like that:
Вложение:
Комментарий к файлу: Выделенный по трём маскам ч.б. растр после размытия и усиления резкости
Dif_and_Unsharp.jpg [ 53.79 Кб | Просмотров: 14094 ]
The fifth step – boundary specifying between black and white.
Вложение:
Комментарий к файлу: Движки исходного диапазона должны быть совмещены. На рисунке они слегка раздвинуты для наглядности.
Levels.jpg [ 45.82 Кб | Просмотров: 14091 ]
Transformation into black-and-white again, and vectorizing.
On the whole, preparing of a B&W image consists of:
Вложение:
Комментарий к файлу: Стадии процесса
Steps.jpg [ 179.17 Кб | Просмотров: 14091 ]
Miscalculation at mask forming for light tints is clear there.
Well, I’m too lazy to do it anew.
There may be more than 3 masks of course. But anyway, it is better
to consider the image first and to make masks for all main cases
of color overlapping than to poke it unthinkingly…
We proceed to vectorizing. Version 7.99 does not provide many
choices here. Nevertheless, it may be reasonable to play with options.
(vectorizing-> appraisal ->Undo->parameter improvement->vectorizing)
Вложение:
Комментарий к файлу: Результат трассировки
Trace.jpg [ 184.32 Кб | Просмотров: 14095 ]
Breakup joining.
We do not recommend to sew lines across gaps just after autotracing.
Clean vector data with Editor a little first. Unfortunately, it is rather
boring operation in 7.99 – I want full-blown Vector Eraser badly...
Вложение:
Комментарий к файлу: Ни какой коррекции линий после трассировки не выполнялось. Сшивка "по живому".
Join.jpg [ 171.59 Кб | Просмотров: 14092 ]
FOR ATTENTION OF BEGINNERS!
Vectorizing technique in Easy Trace supposes
line smoothing and line form optimization
AT THE FINAL STAGE OF PROCESSING.
It means, after final manual correction
of line form and connectedness.
This example demonstrates extraction of
IMAGE lines and does not comprise
Final stages of vectorizing!
Line form optimization see in videos about 8.7P.S. Here is the project for those who wish to train.