Bidirectional Segmented Detection of Land Use Change Based on Object-Level Multivariate Time Series

Abstract

High-precision information regarding the location, time, and type of land use change is integral to understanding global changes. Time series (TS) analysis of remote sensing images is a powerful method for land use change detection. To address the complexity of sample selection and the salt-and-pepper noise of pixels, we propose a bidirectional segmented detection (BSD) method based on object-level, multivariate TS, that detects the type and time of land use change from Landsat images. In the proposed method, based on the multiresolution segmentation of objects, three dimensions of object-level TS are constructed using the median of the following indices: the normalized difference vegetation index (NDVI), the normalized difference built index (NDBI), and the modified normalized difference water index (MNDWI). Then, BSD with forward and backward detection is performed on the segmented objects to identify the types and times of land use change. Experimental results indicate that the proposed BSD method effectively detects the type and time of land use change with an overall accuracy of 90.49% and a Kappa coefficient of 0.86. It was also observed that the median value of a segmented object is more representative than the commonly used mean value. In addition, compared with traditional methods such as LandTrendr, the proposed method is competitive in terms of time efficiency and accuracy. Thus, the BSD method can promote efficient and accurate land use change detection.

Semi-Automatic Methodology for Fire Break Maintenance Operations Detection with Sentinel-2 Imagery and Artificial Neural Network

“…This new high spatial resolution allowed the analysis of the FB conditions, which in current work, was to detect when maintenance operations are performed. It is common to divide remote sensing applications into two groups: land cover classification [7][8][9][10] and change detection [11][12][13][14][15][16][17]. The proposed methodology fits into the second group.…”

“…In Hamunyela et al [12], forest disturbances were detected with resource to two observations and spatio-temporal features and in Hermosilla et al [13]; annual composites were generated to detect changes. In [15][16][17], different approaches were used to identify changes in the land cover and the kind of changes. In [16,17], pixel-based methods were implemented, with three and six observations, respectively, while in [15], an object-based technique was used.…”

“…In [15][16][17], different approaches were used to identify changes in the land cover and the kind of changes. In [16,17], pixel-based methods were implemented, with three and six observations, respectively, while in [15], an object-based technique was used. It should be noticed that in these works [11][12][13][14][15][16][17] Landsat imagery was used, which means that the best achievable detection period is 16 days (but due to the atmospheric conditions the period is usually worse).…”

“…In [16,17], pixel-based methods were implemented, with three and six observations, respectively, while in [15], an object-based technique was used. It should be noticed that in these works [11][12][13][14][15][16][17] Landsat imagery was used, which means that the best achievable detection period is 16 days (but due to the atmospheric conditions the period is usually worse). Additionally, with the exception of [15], more than one observation of previous data is needed for the detection.…”

“…It should be noticed that in these works [11][12][13][14][15][16][17] Landsat imagery was used, which means that the best achievable detection period is 16 days (but due to the atmospheric conditions the period is usually worse). Additionally, with the exception of [15], more than one observation of previous data is needed for the detection. Usually, the change detection relies on the slope of the time-series and consecutive variations in vegetation indices.…”