@article{fdi:010075553,
  title = {{S}parsity optimization method for slow-moving landslides detection in satellite image time-series},
  author = {{P}ham, {M}. {Q}. and {L}acroix, {P}ascal and {D}oin, {M}. {P}.},
  editor = {},
  language = {{ENG}},
  abstract = {{T}his paper presents a new method based on recent optimization technique to detect slow-moving landslides (<150m/year) in time series of displacement field generated by satellite images. {S}parse optimization is applied simultaneously on the 3-{D} data set in space as well as in time. {T}he proposed method takes into account the distinctive signal physical properties in space and time, by enforcing a sparse representation by blocks in space, but a continuing and monotonous representation in time of the landslides. {A}s a result, we show that a mixed l(1,2)-norm is the most suitable norm for this detection problem, compared to pure l(1)-norm or l(2)-norm. {M}oreover, an outlier estimation step is included that sets apart the {G}aussian noise from locally sparse processing errors in the data. {T}he performance of this approach is tested by applying it both on synthetic data and on a time series of displacements fields over 16 dates in the {C}olca {V}alley, {P}eru. {T}his detection presents commission and omission errors for landslides of 29% and 14%, respectively, using a medium resolution (10 m) data set of optical satellite images. {I}t detects all important landslides, already known from field investigations. {M}oreover, it also points out other smaller or unknown landslides, increasing the existing slow-moving landslide inventory by +50%.},
  keywords = {{D}etection ; optimization ; satellite image time-series ; segmentation ; slow-moving landslides ; sparsity},
  booktitle = {},
  journal = {{IEEE} {T}ransactions on {G}eoscience and {R}emote {S}ensing},
  volume = {57},
  numero = {4},
  pages = {2133--2144},
  ISSN = {0196-2892},
  year = {2019},
  DOI = {10.1109/tgrs.2018.2871550},
  URL = {https://www.documentation.ird.fr/hor/fdi:010075553},
}