7 August 2017 Method of image texture segmentation using Laws' energy measures
Author Affiliations +
Proceedings Volume 10445, Photonics Applications in Astronomy, Communications, Industry, and High Energy Physics Experiments 2017; 1044561 (2017) https://doi.org/10.1117/12.2280891
Event: Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments 2017, 2017, Wilga, Poland
Abstract
This article suggests the method of image segmentation, using Laws` texture energy measures. This method allows identifying segments of images efficiently for their further use in the image processing. Laws` measures describe the image most accurately, resulting in making it easier and more efficient in comparison with the other approaches to allocate separate classes of textures. In order to obtain these measures the sixteen masks are calculated. Resulting energy measures can be provided after applying each of the masks to the image. The developed algorithm was tested using a set of test images. Analysis of the obtained results has showed that in case of visually similar texture images the transition to energy maps significantly improves the correlation coefficient and therefore emphasizes textural features of the images and makes it possible to identify the similarities of textures. In order to evaluate the results of efficiency of developed algorithm properly, its results have been compared to the segmentation method based on matrix matches. It was proved that segmentation based on Laws` measures can detect various types of texture more precisely and with greater speed of operation.
© (2017) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Roman Kvyetnyy, Olga Sofina, Alla Olesenko, Paweł Komada, Jan Sikora, Aliya Kalizhanova, Saule Smailova, "Method of image texture segmentation using Laws' energy measures", Proc. SPIE 10445, Photonics Applications in Astronomy, Communications, Industry, and High Energy Physics Experiments 2017, 1044561 (7 August 2017); doi: 10.1117/12.2280891; https://doi.org/10.1117/12.2280891
PROCEEDINGS
9 PAGES


SHARE
Back to Top