Detecting Musk Thistle (Carduus nutans) Infestation Using a Target Recognition Algorithm

作     者：Mustafa Mirik Yves Emendack Ahmed Attia Sriroop Chaudhuri Mimi Roy Georges F. Backoulou Song Cui 

作者机构：Department of Crop Soil and Environmental Sciences Auburn University Auburn USA Middle Tennessee State University Murfreesboro USA Oklahoma State University 127 Noble Research Center Stillwater USA Texas A & M AgriLife Research and Extension Center Vernon USA 

出 版 物：《Advances in Remote Sensing》 (遥感技术进展（英文）)

年 卷 期：2014年第3卷第3期

页      面：95-105页

学科分类：1002[医学-临床医学] 100214[医学-肿瘤学] 10[医学] 

主　　题：Accuracy Assessment Invasive Plant Weed Management Weed Infestation Remote Sensing Geospatial Data Nodding Thistle 

摘      要：The outbreaks of invasive plant species can cause great ecological and agronomic problems through aggressively competing for environmental resources that could be otherwise utilized by other desirable species. Thus, it is crucial for detecting small infestations before they reach a significant extent that can cause ecological and economic damages over a large geological area. Remote sensing is a proven method for mapping invasion extent and pattern based on geospatial imagery and indicated great repeatability, large coverage area, and lower cost compared with traditional ground-based methods before. We investigated the feasibility and performances of adopting multispectral satellite imagery analyses for mapping infestation of musk thistle (Carduus nutans) on native grassland, crop field, and residential areas in early June using spectral angle mapper classifier. Our results showed an overall classification accuracy of 94.5%, indicating great potential of using moderate resolution multispectral satellite-based remote sensing techniques for musk thistle detection over a large spatial scale.