A medical image color correction method based on supervised color constancy.
- Author:
Jiatuo XU
1
;
Liping TU
;
Zhifeng ZHANG
;
Changle ZHOU
Author Information
1. Shanghai University of Traditional Chinese Medicine, Shanghai 201203, China. xjt@fudan.edu.cn
- Publication Type:Journal Article
- MeSH:
Artificial Intelligence;
Calibration;
Color;
standards;
Humans;
Image Processing, Computer-Assisted;
methods;
Pattern Recognition, Automated;
methods;
Photography;
instrumentation;
methods
- From:
Journal of Biomedical Engineering
2010;27(4):721-726
- CountryChina
- Language:Chinese
-
Abstract:
This paper presents a medical image acquisition and analysis method-TRM (Topology Resolve-Map) Model-under natural light condition indoors. Firstly, in accordance to medical image color characteristics, a colorful and grayscale color control patch was made for use as supervised color. "Topology Resolve-Map-Restoration" was carried on in LAB color space of the one-dimensional L* space and the two-dimensional a* b* space. Then, L* value was regulated by subsection regulation and a* b* value was regulated by triangulation topological cutting--close in on center of gravity method. After correction of the 198 color blocks in 22 pictures, the results showed that, by comparison with the standard value, the deltaL*, deltaC* and deltaE decreased significantly (P < 0.01) after correction by TRM. After correction, the difference in image's color is reduced, the color saturation is improved and the value is closer to true value. TRM model can significantly reduce the color difference of the medical image under natural light condition; it has a good effect on color correction.
