Sequential classification of MODIS time series

Grobler, T. L.; Ackermann, Etienne; Zyl, Gusti van; Kleynhans, W.; Salmon, B.P.; Olivier, J.C.

doi:10.1109/igarss.2012.6352683

Cited by 4 publications

(4 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Instead of using traditional likelihood ratios, we exploited the usefulness of relative density ratios estimated directly using RULSIF algorithm as proposed in [27], in deriving the RSPRT statistics. Our framework slightly reduces the correlation in the parameter time series, and unlike CUSUM formulation in [13] and [41], deals with the no-change samples as identically distributed, which is an important assumption of CUSUM. We tested the framework on three different datasets, against different noise level, and also performed cross-validation.…”

Section: Discussionmentioning

confidence: 99%

“…Both the training and testing phases of the proposed framework have been summarized in Algorithms 1 and 2, respectively. It is worth noting here that the authors of [13] and [41] mentioned that both independent and identically distributed (i.i.d.) assumptions were not met in their formulation of CUSUM.…”

Section: A Rsprt With Relative Density Ratio Estimation (M1)mentioning

confidence: 99%

“…After removing the seasonality, the unchanged μ t parameter time series has slightly reduced correlation but not enough to be considered as negligible. However, our formulation considers all the no-change samples as coming from a single distribution, unlike CUSUM formulation in [13], [41]. Therefore, the no-change samples can be considered as identically distributed, and the change is detected when this assumption is violated, i.e., when a sample from a significantly different distribution is encountered.…”

Section: A Rsprt With Relative Density Ratio Estimation (M1)mentioning

confidence: 99%

See 2 more Smart Citations

A Relative Density Ratio-Based Framework for Detection of Land Cover Changes in MODIS NDVI Time Series

Anees

Aryal

O’Reilly

et al. 2016

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

To improve statistical approaches for near real-time land cover change detection in nonGaussian time-series data, we propose a supervised land cover change detection framework in which a MODIS NDVI time series is modeled as a triply modulated cosine function using the extended Kalman filter and the trend parameter of the triply modulated cosine function is used to derive repeated sequential probability ratio test (RSPRT) statistics. The statistics are based on relative density ratios estimated directly from the training set by a relative unconstrained least squares importance Fitting (RULSIF) algorithm, unlike traditional likelihood ratio-based test statistics. We test the framework on simulated, synthetic, and real-world beetle infestation datasets, and show that using estimated relative density ratios, instead of assuming the individual density functions to be Gaussian or approximating them with Gaussian Kernels, in the RSPRT statistics achieves better performance in terms of accuracy and detection delay. We verify the efficiency of the proposed approach by comparing its performance with three existing methods on all the three datasets under consideration in this study. We also propose a simple heuristic technique that tunes the threshold efficiently in difficult cases of near real-time change detection, when we need to take three performance indices, namely, false positives, false negatives, and mean detection delay, into account simultaneously. Index Terms-Change detection, extended Kalman filter (EKF), model fitting, MODIS, relative density ratio, time series. I. INTRODUCTIONL AND COVER change detection research has seen significant recent contributions [1]- [19]. However, every proposed framework has its limitations on global scale due to particularity of the task at hand and the circumstances under which it is developed. Hence, no single framework is optimal in a wide range of scenarios simultaneously. Therefore, an efficient change detection framework is always in demand for a particular task and circumstances under consideration. In this Manuscript

show abstract

Section: Discussionmentioning

confidence: 99%

Section: A Rsprt With Relative Density Ratio Estimation (M1)mentioning

confidence: 99%

Section: A Rsprt With Relative Density Ratio Estimation (M1)mentioning

confidence: 99%

See 1 more Smart Citation

A Relative Density Ratio-Based Framework for Detection of Land Cover Changes in MODIS NDVI Time Series

Anees

Aryal

O’Reilly

et al. 2016

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Focusing on describing ocean regions with their temporal characteristics, and clarifying what they represent, would enable subsequent research to exploit more targeted supervised approaches. Research using unsupervised methodologies, or adapted methodologies, has included the use of ISODATA clustering [13,20,28,30], k-means clustering [49,55,76], minimum error [77], maximum likelihood [78], Ward's method [55], and expectation maximisation [55]. k-means and ISODATA clustering are by far the more commonly used approaches.…”

Section: Classification (Cls)-founded Approachesmentioning

confidence: 99%

From Land to Sea, a Review of Hypertemporal Remote Sensing Advances to Support Ocean Surface Science

Scarrott

Cawkwell

Jessopp

et al. 2019

Water

View full text Add to dashboard Cite

Increases in the temporal frequency of satellite-derived imagery mean a greater diversity of ocean surface features can be studied, modelled, and understood. The ongoing temporal data “explosion” is a valuable resource, having prompted the development of adapted and new methodologies to extract information from hypertemporal datasets. Current suitable methodologies for use in hypertemporal ocean surface studies include using pixel-centred measurement analyses (PMA), classification analyses (CLS), and principal components analyses (PCA). These require limited prior knowledge of the system being measured. Time-series analyses (TSA) are also promising, though they require more expert knowledge which may be unavailable. Full use of this resource by ocean and fisheries researchers is restrained by limitations in knowledge on the regional to sub-regional spatiotemporal characteristics of the ocean surface. To lay the foundations for more expert, knowledge-driven research, temporal signatures and temporal baselines need to be identified and quantified in large datasets. There is an opportunity for data-driven hypertemporal methodologies. This review examines nearly 25 years of advances in exploratory hypertemporal research, and how methodologies developed for terrestrial research should be adapted when tasked towards ocean applications. It highlights research gaps which impede methodology transfer, and suggests achievable research areas to be addressed as short-term priorities.

show abstract

Retrieval and validation of long-term aerosol optical depth from AVHRR data over China

Jin

Xue

Jiang

et al. 2022

International Journal of Digital Earth

View full text Add to dashboard Cite

Sequential classification of MODIS time series

Cited by 4 publications

References 3 publications

A Relative Density Ratio-Based Framework for Detection of Land Cover Changes in MODIS NDVI Time Series

A Relative Density Ratio-Based Framework for Detection of Land Cover Changes in MODIS NDVI Time Series

From Land to Sea, a Review of Hypertemporal Remote Sensing Advances to Support Ocean Surface Science

Retrieval and validation of long-term aerosol optical depth from AVHRR data over China

Contact Info

Product

Resources

About