OIC-MCE: A Practical Land Cover Mapping Approach for Limited Samples Based on Multiple Classifier Ensemble and Iterative Classification

Abstract

Land cover samples are usually the foundation for supervised classification. Unfortunately, for land cover mapping in large areas, only limited samples can be used due to the time-consuming and labor-intensive sample collection. A novel and practical Object-oriented Iterative Classification method based on Multiple Classifiers Ensemble (OIC-MCE) was proposed in this paper. It systematically integrated object-oriented segmentation, Multiple Classifier Ensemble (MCE), and Iterative Classification (IC). In this method, the initial training samples were updated self-adaptively during the iterative processes. Based on these updated training samples, the inconsistent regions (ICR) in the classification results of the MCE method were reclassified to reduce their uncertainty. Three typical case studies in the China-Pakistan Economic Corridor (CPEC) indicate that the overall accuracy of the OIC-MCE method is significantly higher than that of the single classifier. After five iterations, the overall accuracy of the OIC-MCE approach increased by 5.58%-8.38% compared to the accuracy of the traditional MCE method. The spatial distribution of newly added training samples generated by the OIC-MCE approach was relatively uniform. It was confirmed by ten repeated experiments that the OIC-MCE approach has good stability. More importantly, even if the initial sample size reduced by 65%, the quality of the final classification result based on the proposed OIC-MCE approach would not be greatly affected. Therefore, the proposed OIC-MCE approach provides a new solution for land cover mapping with limited samples. Certainly, it is also well suited for land cover mapping with abundant samples.Remote Sens. 2020, 12, 987 2 of 21 various homogeneous or heterogeneous data sources, advanced machine learning algorithms, and a variety of sampling and classification strategies have been introduced into land cover mapping researches [1,[5][6][7].Supervised classification is the most commonly used method in land cover mapping studies, especially for generating a regional or even a global land cover product [8]. Samples are the foundation of this method. The quantity and quality of samples would have a significant impact on the accuracy of land cover products [9,10]. There are many methods to collect samples, such as field survey, crowdsourcing, and manual interpretation [11]. In general, field survey has been a simple and practical method, but it cannot be conducted in inaccessible regions [12]. More importantly, it is time-consuming and labor-intensive. Manual interpretation of high-resolution remote sensing images, relying on the interpretation experiences of the interpreters, has been an effective way to fill the blank of field surveys. However, the subjectivity of the interpreter might affect the objectivity of samples [10]. Crowdsourcing was a new technology that helps to collect samples [13,14]. Unfortunately, it is still in the experimental stage at present, with few available samples and a lack of objective evaluation for their reliability...

Multitemporal Relearning With Convolutional LSTM Models for Land Use Classification

“…Furthermore, Shi et al [31] developed an active relearning framework that can improve the classification results with less labelling costs. More recently, Lei et al [32] developed an object-oriented classification method which iteratively integrates classification results. The experimental results achieved promising accuracy even with a limited number of samples.…”

A review of land use/land cover change mapping in the China-Central Asia-West Asia economic corridor countries

“…Given the fact that the annual percentage of LC and LU can barely exceed a few percent of the total land area (Chen et al, 2019), migration of reference data from a specific time (year) to another time (target year) has been recently implemented to address the lack of accurate and reliable data (Ghorbanian et al, 2020). The crowdsourcing technology has been also proposed to help to sample collection in recent years (Lei et al, 2020).…”

“…However, there are still many issues related to LULCC studies that need more attention from scholars. The lack of a standard assessment system (Verburg, Alexander, Evans, Magliocca, & Vliet, 2019;Ye, Pontius, & Rakshit, 2018) and lack of reliable and sufficient ground truth data (Lei et al, 2020) are two important examples in this regard. Furthermore, with the leaps in Remote Sensing (RS) sensors and platforms along with the advent of new concepts, such as cloud computing, RS Big Data has entered into the field of LULCC monitoring, which demands much more attention from scholars (Comber & Wulder, 2019;Xing & Sieber, 2016).…”

Progress in Grassland Cover Conservation in Southern European Mountains by 2020: A Transboundary Assessment in the Iberian Peninsula with Satellite Observations (2002–2019)

“…That provides the opportunity to integrate independent classification results of different algorithms (e.g., those based on artificial intelligence, such as random forests, k-nearest neighbour, decision trees) through ensemble strategies (e.g., soft voting). Improvement of the accuracy of classification results compared to single classifiers has been observed [28,29]. MCE also yields class-allocation probability layers at the pixel level, which can be used to generate class-allocation uncertainty information (e.g., Shannon entropy, [30]) and support refinement operations before map comparisons, reporting pixel assignment confidence and analyses of the relationship between class-allocation uncertainty and classification accuracy.…”

“…Grasslands of the study area are representative of ancient mountain pastoral systems of the Iberian Peninsula, which are rapidly transforming with farmland change and abandonment, and are a target of European policies [13]. Specifically, we used Landsat imagery and derivatives within a multiple classifier ensemble approach [28] to map grassland cover in 2002 and 2019. Post classification comparison of uncertainty refined grassland cover maps to identify changed areas.…”