A novel built-up spectral index developed by using multiobjective particle-swarm-optimization technique

8th IGRSM International Conference and Exhibition on Geospatial & Remote Sensing

地球科学;计算机科学

Sameen, Maher Ibrahim^1 ; Pradhan, Biswajeet^1

Department of Civil Engineering, Faculty of Engineering, University Putra Malaysia, Selangor

43400, Malaysia^1

关键词: Accuracy assessment;    Built-up areas;    Classification accuracy;    Classification results;    Development process;    Multi objective particle swarm optimization;    Particle swarm optimization technique;    Spectral indices;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/37/1/012006/pdf
DOI  :  10.1088/1755-1315/37/1/012006

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

In this study, we propose a novel built-up spectral index which was developed by using particle-swarm-optimization (PSO) technique for Worldview-2 images. PSO was used to select the relevant bands from the eight (8) spectral bands of Worldview-2 image and then were used for index development. Multiobiective optimization was used to minimize the number of selected spectral bands and to maximize the classification accuracy. The results showed that the most important and relevant spectral bands among the eight (8) bands for built-up area extraction are band4 (yellow) and band7 (NIR1). Using those relevant spectral bands, the final spectral index was form ulated by developing a normalized band ratio. The validation of the classification result using the proposed spectral index showed that our novel spectral index performs well compared to the existing WV -BI index. The accuracy assessment showed that the new proposed spectral index could extract built-up areas from Worldview-2 image with an area under curve (AUC) of (0.76) indicating the effectiveness of the developed spectral index. Further improvement could be done by using several datasets during the index development process to ensure the transferability of the index to other datasets and study areas.