Object-Based Classification by using Hierarchical Segmentation and Weighted Genetic Algorithm
Hyperspectral imaging concerns measurement and interpretation of spectral imagery acquired by satellite, airborne, terrestrial, or laboratory sensors over visible, infrared and sometime thermal spectral regions of electromagnetic spectrum. There are two major approaches for classification of hyperspectral images: the spectral or pixel-based techniques, and the spectral-spatial or object-based techniques. Recently, an effective approach for spectral-spatial classification has been proposed using Hierarchical SEGmentation (HSEG) grown form automatically selected markers. This paper aims at improving this approach for classification of hyperspectral images in urban areas. The Weighted Genetic (WG) algorithm is first used to obtain the subspace of hyperspectral data. The obtained features are then fed into the marker-based HSEG algorithm. Then, the contextual features from segmented images are extracted. For spatial features, area, entropy, shape, adjacency and relation features are considered as the potential components in feature space. Finally, using both spectral and spatial features, the image objects are classified by a rule-based classifier. The experimental tests are applied to two datasets: the Berlin, and Quebec City, which are two known and benchmark datasets in hyperspectral imagery. The evaluation of results showed that the proposed approach achieves approximately 16% and 9% better overall accuracy than the Original-HSEG algorithm for these datasets respectively.
- حق عضویت دریافتی صرف حمایت از نشریات عضو و نگهداری، تکمیل و توسعه مگیران میشود.
- پرداخت حق اشتراک و دانلود مقالات اجازه بازنشر آن در سایر رسانههای چاپی و دیجیتال را به کاربر نمیدهد.