Title :
Natural Image Segmentation Based on Precise Edge Detection
Author :
Wenya Feng ; Yilin Guo ; Xiaoyu Shi ; Yonggan Hou
Abstract :
Based on a new variational-based model within the fuzzy framework, we propose a new solution to the problem of multi-region segmentation of natural images. The advantages of our model is: by introducing the PCA features and modeling regions by Gaussian distribution, the proposed model can partition texture images better than classical variational-based segmentation models. We use the Berkeley database(BSDS300) as sample source and compare this model with some other well-known models such as the level-set model and the fuzzy region competition model. Comprehensive experiments have proven that the proposed model has better segmentation performance and faster speed.
Keywords :
Gaussian distribution; edge detection; image segmentation; natural scenes; principal component analysis; BSDS300; Berkeley database; Gaussian distribution; PCA features; fuzzy framework; fuzzy region competition model; level-set model; multiregion segmentation; natural image segmentation; precise edge detection; texture images; variational-based model; Hidden Markov models; Image color analysis; Image edge detection; Image segmentation; Lighting; Numerical models; Principal component analysis; Gaussian mixture models; Natural image segmentation; PCA; Variational model;
Conference_Titel :
Machine Learning and Applications (ICMLA), 2013 12th International Conference on
Conference_Location :
Miami, FL
DOI :
10.1109/ICMLA.2013.47