Texture detection in noisy images by combining several local parameters
Alexey V. Naumenko, Sergey Krivenko, Nikolay N. Ponomarenko, Alexander A. Zelensky, Владимир Васильевич Лукин · 2015
A problem of detecting textural areas in images corrupted by noise is considered. Detection is based on joint use of several local parameters calculated in scanning windows (blocks) of different size. Trained support vector machine (SVM) classifier is used for combining local parameters. Factors that influence detector performance are analyzed. It is shown that detector performance can be improved by taking into account information from classifier output for neighbor pixels.