Kirzieiev A. A., Matsuga O. N., Bondar O. E.
Oles Honchar Dnipropetrovsk National University
THE SOFTWARE OF IMAGE COLORIZATION
Until recently, the opportunities of photographic studios were limited by the equipment which could take photos only in a black-and-white format. However, over time, this limitation was overcome successfully. The modern equipment has come instead of the old one. It became the impulse to develop a new trend such as colorization.
Colorization is a computer process where an original black-and-white image is converted into a colored one with a particular algorithm. There are many such algorithms, among which the most common are DrawCut, Levin’s algorithm, Yuvin Taya’s algorithm, etc.
The software was developed in NetBeans 7.4 environment in Java programming language with a view to show colorization process of images. The mathematical apparatus is the algorithm which has been described in the article . The program named «ImageColorization», the interface of which is shown in Figure, provides the following features:
1) a wide choice of color palette;
2) width adjustment of the brush;
3) an ability to cancel or return to the last 5 user actions;
4) interactive conversion of a black-and-white photo into a colored one;
5) an ability to refine the resulting image;
6) the resulting image can be saved in different formats such as *. bmp, *. jpg, *. png;
7) the availability of a user manual.
As a result, the proposed program has a set of the above-mentioned features and the advantage offered by Java – the independence of the application from the installed operating system.
Fig. 1. Program interface
The main drawbacks of the software can be caused by the defects of the algorithm used. In particular, the colorization of large images takes quite a long time (for example, the coloring image of size 1900x1200 pixels took 29 minutes and 37 seconds on the CPU AMD Athlon (tm) II Dual-Core M320 2.10 GHz with 2048 RAM). Another disadvantage can be seen in an imperfect blending of colors on the borders of objects. Therefore, this product should be used for images with resolution up to size 800x800 pixels and is recommended for the coloring which is equidistant from the borders of an object (as in Fig.).
The list of references:
1. Konushin V. Interactive Image Colorization and Recoloring based on Coupled Map Lattices [Electronic resource] / V. Konushin, V. Vezhnevets. – Mode of access: http://www.graphics.cs.msu.ru/en/publications/text/gc2006kv.pdf