Self-Learning Based Land-Cover Classification Using Sequential Class Patterns from Past Land-Cover Maps

Abstract

Abstract:To improve the accuracy of classification with a small amount of training data, this paper presents a self-learning approach that defines class labels from sequential patterns using a series of past land-cover maps. By stacking past land-cover maps, unique sequence rule information from sequential change patterns of land-covers is first generated, and a rule-based class label image is then prepared for a given time. After the most informative pixels with high uncertainty are selected from the initial classification, rule-based class labels are assigned to the selected pixels. These newly labeled pixels are added to training data, which then undergo an iterative classification process until a stopping criterion is reached. Time-series MODIS NDVI data sets and cropland data layers (CDLs) from the past five years are used for the classification of various crop types in Kansas. From the experiment results, it is found that once the rule-based labels are derived from past CDLs, the labeled informative pixels could be properly defined without analyst intervention. Regardless of different combinations of past CDLs, adding these labeled informative pixels to training data increased classification accuracy and the maximum improvement of 8.34 percentage points in overall accuracy was achieved when using three CDLs, compared to the initial classification result using a small amount of training data. Using more than three consecutive CDLs showed slightly better classification accuracy than when using two CDLs (minimum and maximum increases were 1.56 and 2.82 percentage points, respectively). From a practical viewpoint, using three or four CDLs was the best choice for this study area. Based on these experiment results, the presented approach could be applied effectively to areas with insufficient training data but access to past land-cover maps. However, further consideration should be given to select the optimal number of past land-cover maps and reduce the impact of errors of rule-based labels.

Impact of Texture Information on Crop Classification with Machine Learning and UAV Images

“…Fourteen texture features defined by Haralick et al [20] are correlated, indicating that using all possible texture features provides redundant spatial contextual information which is not useful for classification. In this study, six texture features [46] were considered: (1) mean (ME), (2) standard deviation (STD), (3) homogeneity (HOM), (4) dissimilarity (DIS), (5) entropy (ENT), and (6) angular second moment (ASM), presented in Equations (1) to (6):…”

“…Agricultural environments are known to be sensitive to abnormal weather conditions and climatic disasters such as drought and flood [1,2], thus rendering essential systematic monitoring of crop conditions and crop yield forecasting [3,4]. Remote sensing technology received attention in the agriculture community due to its ability to provide periodic and regional information for crop monitoring and thematic mapping [5,6].…”

“…Crop type maps derived from classification of remote sensing images are important resources for crop yield estimation and forecasting. Since any error in the crop type maps affects outputs of crop yield and forecasting models, it is critical to generate reliable crop type maps [6]. The most important elements of input remote sensing images for crop classification are their spatial and temporal resolutions.…”

Impact of the Urbanization on Coastal Dune: Case of Kharrouba, West of Algeria

“…The Algerian Ministry of Land-use Planning, Tourism, and Handicraft (2015) mentioned that the available host accommodation in 2014 could satisfy the demand of 11000 tourists whereas the Province of Mostaganem registered in the same year 10 million visitors during the summer. Consequently, the inadequacy between tourism demand and offer would probably lead the decision maker to build more host structure close to the shore, that would mean a progressive dune destruction in favor to the urbanization (Paskoff, 1994;Scarelli et al, 2017;Kim et al, 2017). In this period (2012-2016), two breakwaters have been constructed with the aim to protect the beach of Sidi Medjdoub, the most popular bathing site of Kharrouba, against the marine erosion; according to Rizzi et al (2015), such a protection presented a weak resilience against storms and rainfalls.…”

Semantic Segmentation Based Remote Sensing Data Fusion on Crops Detection

“…At present RS technology has received great attention in the agriculture community due to its ability to provide periodic and regional information for crop monitoring and thematic mapping [1] [2]. Modern RS to identify any features on the surface is no longer considered as a processing of a one-source single date image.…”