Object-based crop classification in Hetao plain using random forest
- PDF / 3,826,855 Bytes
- 13 Pages / 595.276 x 790.866 pts Page_size
- 43 Downloads / 234 Views
RESEARCH ARTICLE
Object-based crop classification in Hetao plain using random forest Tengfei Su 1 & Shengwei Zhang 1,2 Received: 25 November 2019 / Accepted: 25 September 2020 # Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Crop classification based on object-based image analysis (OBIA) is increasingly reported. However, it is still challenging to produce high-quality crop type maps by using recent techniques. This article introduces a new object-based crop classification algorithm which contains 4 steps. First, a random forest (RF) classifier is trained by using the initial training set, which tends to have a relatively small size. Second, importance scores for each feature variable are derived by using the RF model. Third, by treating the importance scores as weighting factors, a weighted Euclidean distance criterion is designed and used for sample creation to enlarge training set. Fourth, RF is re-trained by using the enlarged training set, and then it is employed for final classification. To validate the proposed strategy, a Worldview-2 image covering a part of Hetao plain is experimented. Results indicate that the new method yields the best overall accuracy, which equals 90.52%. Keywords Object-based image analysis . Crop classification . Random forest
Introduction The work of agricultural statistics plays an indispensible role in almost all of the governments throughout the world. An important part of this operation is to obtain acreage information of different crops (Yan and Roy 2014; Boryan et al. 2011; Wardlow and Egbert 2010). To finish this job, traditional approaches are mainly based on on-site visiting and sampling, which severely suffers from high labor cost and inevitable error rate (Wardlow and Egbert 2010). To solve this issue, remote sensing has been widely employed for this task, since it is deemed as an efficient tool for large scale surveying and mapping. According to a recent report of the provincial government of Inner Mongolia of China, in the year of 2017, 8977 scenes of remotely sensed images were processed and analyzed for the extraction of crop field information in this
Communicated by: H. Babaie * Shengwei Zhang [email protected] 1
College of Water Conservancy and Civil Engineering, Inner Mongolia Agricultural University, No. 306, Zhaowuda Road, Hohhot, China
2
Inner Mongolia Autonomous Region Key Laboratory of Big Data Research and Application of Agriculture and Animal Husbandry, No. 306, Zhaowuda Road, Hohhot, China
province (www.nmagri.gov.cn/). This example implies that remote sensing is a valuable asset for agricultural sectors. The increasing applications of agricultural remote sensing have attracted a great number of researchers who have developed new models and methodologies for crop field mapping. Among these efforts, much attention has been paid to objectbased image analysis (OBIA). A possible explanation is that OBIA uses segment instead of pixel as the processing unit, and in agricultural landscape, crop fields appear as contiguous segments, which
Data Loading...
