Background Subtraction using Adaptive Singular Value Decomposition

Reitberger, Günther; Sauer, Tomas

doi:10.1007/s10851-020-00967-4

Cited by 10 publications

(9 citation statements)

References 31 publications

(47 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…A classic anomaly detection strategy consists in performing a background subtraction -for example with a principal component analysis (PCA) (Reitberger andSauer, 2019, Ouerghi et al, 2021) -then locally modeling the remaining residual image as stationary Gaussian. This probabilistic model then enables Neyman-Pearson tests to be carried out to detect anomalous pixels (Manolakis andShaw, 2002, Matteoli et al, 2010).…”

Section: Introductionmentioning

confidence: 99%

Automatic Methane Plumes Detection in Time Series of Sentinel-5p L1b Images

Ouerghi

Ehret

Franchis

et al. 2022

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. Reducing methane emissions is essential to tackle climate change. Here, we address the problem of detecting automatically large methane leaks using hyperspectral data from the Level 1B product of the Sentinel-5P satellite. To do this, two features of TROPOMI (TROPOspheric Monitoring Instrument), the Sentinel-5P satellite sensor, are exploited. The first one is the fine spectral sampling of the data which allows to isolate features of the methane absorption spectrum in the shortwave infrared wavelength range (SWIR). The second one is the daily coverage of the whole Earth which allows to perform time series analysis. Our method involves three main steps: i) a pixel reconstruction, ii) an angle correction and iii) a plume detection with a time series. In the first step, a simplified absorption model is inverted to recover, for each pixel, a coefficient representative of the presence of methane which we call the methane coefficient. In the second step, a correction is made to the methane coefficient to take into account the viewing angle of the satellite. In the third step, the obtained coefficient is normalized spatially and then the detection is carried out pixel by pixel, by looking for anomalies in a time series. We validate our method by comparing the detected plumes against a recently published dataset of plumes manually detected in the Sentinel-5P L2 methane product. We then show how our method can complement the Sentinel-5P L2 methane product for the detection of methane plumes.

show abstract

Section: Introductionmentioning

confidence: 99%

Automatic Methane Plumes Detection in Time Series of Sentinel-5p L1b Images

Ouerghi

Ehret

Franchis

et al. 2022

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

show abstract

“…Although the absorption features of CH4 and water vapor (H2O) sometimes overlap in the infrared spectrum, their separation can be addressed by a wise selection of wavelengths (Crevoisier et al, 2009). A classic anomaly detection strategy consists in performing a background subtraction -for example with a principal component analysis (PCA) (Reitberger and Sauer, 2019) -then locally modeling the remaining residual image as following a Gaussian model. This probabilistic model then enables Neyman-Pearson tests to be carried out on the pixels to detect anomalies (Manolakis andShaw, 2002, Matteoli et al, 2010).…”

Section: Introductionmentioning

confidence: 99%

Detection of Methane Plumes in Hyperspectral Images From Sentinel-5p by Coupling Anomaly Detection and Pattern Recognition

Ouerghi

Ehret

Franchis

et al. 2021

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. Reducing methane emissions is essential to tackle climate change. Here, we address the problem of detecting large methane leaks using hyperspectral data from the Sentinel-5P satellite. For that we exploit the fine spectral sampling of Sentinel-5P data to detect methane absorption features visible in the shortwave infrared wavelength range (SWIR). Our method involves three separate steps: i) background subtraction, ii) detection of local maxima in the negative logarithmic spectrum of each pixel and iii) anomaly detection in the background-free image. In the first step, we remove the impact of the albedo using albedo maps and the impact of the atmosphere by using a principal component analysis (PCA) over a time series of past observations. In the second step, we count for each pixel the number of local maxima that correspond to a subset of local maxima in the methane absorption spectrum. This counting method allows us to set up a statistical a contrario test that controls the false alarm rate of our detections. In the last step we use an anomaly detector to isolate potential methane plumes and we intersect those potential plumes with what have been detected in the second step. This process strongly reduces the number of false alarms. We validate our method by comparing the detected plumes against a dataset of plumes manually annotated on the Sentinel-5P L2 methane product.

show abstract

“…During the past two decades, many methods [ 1 , 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 ] were proposed for the background initialization task. In general, these techniques can be classified into four main categories: pixel-based methods [ 1 , 7 , 8 , 9 , 10 , 11 ], iterative-based methods [ 12 , 13 , 14 , 15 ], low-rank/sparse data separation methods [ 16 , 17 , 18 , 19 , 20 , 21 , 22 ], and deep learning-based methods [ 23 , 24 , 25 , 26 , 27 ].…”

Section: Introductionmentioning

confidence: 99%

“…The second subcategory includes iterative-based methods [ 12 , 13 , 14 , 15 ]. These methods usually consist of two stages.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Singular Spectrum Analysis for Background Initialization with Spatio-Temporal RGB Color Channel Data

Wang

Liang

2021

Entropy

View full text Add to dashboard Cite

In video processing, background initialization aims to obtain a scene without foreground objects. Recently, the background initialization problem has attracted the attention of researchers because of its real-world applications, such as video segmentation, computational photography, video surveillance, etc. However, the background initialization problem is still challenging because of the complex variations in illumination, intermittent motion, camera jitter, shadow, etc. This paper proposes a novel and effective background initialization method using singular spectrum analysis. Firstly, we extract the video’s color frames and split them into RGB color channels. Next, RGB color channels of the video are saved as color channel spatio-temporal data. After decomposing the color channel spatio-temporal data by singular spectrum analysis, we obtain the stable and dynamic components using different eigentriple groups. Our study indicates that the stable component contains a background image and the dynamic component includes the foreground image. Finally, the color background image is reconstructed by merging RGB color channel images obtained by reshaping the stable component data. Experimental results on the public scene background initialization databases show that our proposed method achieves a good color background image compared with state-of-the-art methods.

show abstract

Background Subtraction using Adaptive Singular Value Decomposition

Cited by 10 publications

References 31 publications

Automatic Methane Plumes Detection in Time Series of Sentinel-5p L1b Images

Automatic Methane Plumes Detection in Time Series of Sentinel-5p L1b Images

Detection of Methane Plumes in Hyperspectral Images From Sentinel-5p by Coupling Anomaly Detection and Pattern Recognition

Singular Spectrum Analysis for Background Initialization with Spatio-Temporal RGB Color Channel Data

Contact Info

Product

Resources

About