Title :
Multi-step segmentation of images by means of a recurrent neural network
Author :
Nemirovskiy, V.B. ; Stoyanov, A.K.
Author_Institution :
Computer-Aided Design Systems Department, Tomsk Polytechnic University, Rusia
Abstract :
The application of recurrent neural network for image segmentation is considered. The described method allows to perform clustering of the image parameters without the a priori information about the clusters number. Calculation of network parameters for segmentation is discussed. The experiments results of multi-step segmentation are given.
Keywords :
image segmentation; pattern clustering; recurrent neural nets; cluster number; image parameter clustering; multistep image segmentation; network parameters; recurrent neural network; Arrays; Brightness; Entropy; Image color analysis; Image segmentation; Neurons; Recurrent neural networks; critical point; image; pixel; point map; segmentation;
Conference_Titel :
Strategic Technology (IFOST), 2012 7th International Forum on
Conference_Location :
Tomsk
Print_ISBN :
978-1-4673-1772-6
DOI :
10.1109/IFOST.2012.6357619
