Self-supervised texture segmentation using complementary types of features

✍ Scribed by Jiebo Luo; Andreas E. Savakis

• Publisher: Elsevier Science
• Year: 2001
• Tongue: English
• Weight: 460 KB
• Volume: 34
• Category: Article
• ISSN: 0031-3203
• DOI: 10.1016/s0031-3203(00)00146-1

✦ Synopsis

A two-stage texture segmentation approach is proposed where an initial segmentation map is obtained through unsupervised clustering of multiresolution simultaneous autoregressive (MRSAR) features and is followed by selfsupervised classi"cation of wavelet features. The regions of high con"dencea and low con"dencea are identi"ed based on the MRSAR segmentation result using multilevel morphological erosion. The second-stage classi"er is trained by the high-con"dencea samples and is used to reclassify only the low-con"dencea pixels. The proposed approach leverages on the advantages of both MRSAR and wavelet features. Experimental results show that the misclassi"cation error can be signi"cantly reduced by using complementary types of texture features.