Nach oben pdf A Benchmark Evaluation of Similarity Measures for Multi-temporal SAR Image Change Detection

A Benchmark Evaluation of Similarity Measures for Multi-temporal SAR Image Change Detection

A Benchmark Evaluation of Similarity Measures for Multi-temporal SAR Image Change Detection

Most unsupervised change detection methods comprise three steps: preprocessing such as despeckling and co- registration, image comparison to generate a change index, and change index analysis to generate a final binary change map. Preprocessing includes image registration, radiometric corrections, despeckling, etc. One widely used image com- parison technique for SAR images is the log-ratio operator [4]–[6] which is particularly suited to SAR change detection due to the presence of multiplicative noise. Recently, methods based on information measures have shown promising per- formance for multi-temporal change detection. They assess image similarity by quantifying the dependence or distance between two random variables associated with two images. One prominent work in [7] proposed a method for multi- temporal SAR change detection based on the evolution of the local statistics, which was extended to object-based change detection in [8]. A similar method has been extended to the wavelet domain [9]. In [10], several information similarity measures including distance to independence, mutual infor- mation, cluster reward, Woods criterion, and correlation ratio, were compared for change detection, among which mutual information has been demonstrated to be rather efficient. Tak- ing advantage of mutual information, a pixel-based approach comparing localized mutual information was proposed in [11]. Intuitively, if two pixels share a lot of information, it is reasonable to assume no change at their location. Based on this idea, another information measure for change detection derived from mutual information was introduced in [12], namely mixed information, which unifies mutual information and variational information by a parameter. Furthermore, stochastic kernels including both Kullback-Leibler divergence and mutual information were used in [13] as features in a support vector machine for SAR change detection. Based on the estimation of a bivariate Gamma distribution, mutual information was applied to SAR change detection in [14]. A region-based local mutual information change indicator was proposed by [15] to perform a change analysis of urbanization processes from multi-temporal panchromatic SPOT 5 images. Through a two-scale implementation, mutual information can be split into two terms to be linked to a change detection part and a registration part [16].
Mehr anzeigen

18 Mehr lesen

An Innovative Curvelet-only-Based Approach for Automated Change Detection in Multi-Temporal SAR Imagery

An Innovative Curvelet-only-Based Approach for Automated Change Detection in Multi-Temporal SAR Imagery

Obviously, the real and imaginary parts of the Curvelet coefficients follow more or less a normal distribution with its mean in zero and a similar standard deviation. This was expected because of two facts: At first, the large number of real random values associated with the Curvelet coefficients real and imaginary parts induces compulsorily a normal distribution because—according to the central limit theorem [22]—any distribution tends towards a normal distribution for a sufficiently large number of samples. And secondly, regarding change images in remote sensing applications—where (without exception) only local changes are present—the most part of the image is more or less stochastic, i.e., there are only few distinct structures that possibly affect the normal distribution of the coefficients. Above all, the single sub-band coefficients being normalized during Curvelet transform [23] make a special treatment of the sub-bands dispensable. Thus, if the image contents were completely stochastic, the random combination of real and imaginary parts would result in amplitudes that follow a Rayleigh distribution—similar to SAR amplitudes of distributed targets as reported in [24]—uniformly over all sub-bands. The sole parameter of the Rayleigh distribution function then is equal to the standard deviation of the real and imaginary parts. Figure 8 illustrates the analytical probability density function (PDF), the cumulative density function (CDF), and the empirical histogram of the Curvelet coefficient amplitudes. Indeed the histogram of the coefficient amplitudes does not deviate that much from the Rayleigh PDF except for very low and very high values. On the one hand, there are more low values in the empirical histogram than expected. On the other hand, there are more high amplitudes than the Rayleigh PDF would predict. From this, it follows that the combination of real and imaginary parts is not completely stochastic, but partly deterministic which is caused by the small amount of distinct structures in the change images.
Mehr anzeigen

28 Mehr lesen

Feature based non parametric estimation of Kullback Leibler divergence for SAR image change detection

Feature based non parametric estimation of Kullback Leibler divergence for SAR image change detection

measures have been applied to change detection as well and have shown promising performances. A prominent work by Inglada and Mercier ( 2007 ) proposed a method for multi-temporal SAR change detection based on the evolution of local statistics computed from the pre-event and post-event images. The local statistics are estimated by one-dimen- sional Edgeworth series expansion, which approximates the probability density functions of the pixels in the neighbourhood. The degree of evolution of the local statistics is measured using the Kullback –Leibler divergence. Bovolo and Bruz-zone (2008) extended this method to object-based change detection by computing the Kullback –Leibler divergence of two corre- sponding regions obtained by image segmentation. An unsupervised change detection method in the wavelet domain based on statistical wavelet coe fficient modelling was proposed by Cui and Datcu ( 2012 ). Several information similarity measures, namely distance to independence, mutual information, cluster reward, Woods criterion and correlation ratio, were compared by Alberga ( 2009 ) for change detection, among which mutual information has been demonstrated to be rather e fficient. Taking advantage of mutual information, a pixel-based approach comparing localized mutual information was proposed by Winter et al. ( 1997 ). Intuitively, if two pixels share a lot of information, it is reasonable to assume no change at their location. Based on this idea, another information measure for change detection derived from mutual information was introduced by Gueguen and Datcu ( 2009 ), namely mixed information, which uni fies mutual information and variational information by a para- meter. Furthermore, stochastic kernels including both Kullback –Leibler divergence and mutual information were used by Mercier et al. ( 2006 ) as features in a support vector machine for SAR change detection. Based on the estimation of a bivariate Gamma distribution, mutual information was applied to SAR change detection by Chatelain et al. ( 2007 ). Through a two- scale implementation, mutual information can be split into two terms to be linked to a change detection part and a registration part (Mercier and Inglada 2008 ). The method by Bovolo and Bruzzone ( 2005 ) exploits a wavelet-based multiscale decomposition of the log-ratio image aiming at representation of the change signal using di fferent scales. k-means clustering was applied by Celik ( 2009 ) to classify the undecimated wavelet transform of the di fference image into two classes corresponding to change and no-change classes. A benchmark evaluation of similarity measures, namely mutual information, variational information, mixed information and Kullback –Leibler divergence, was carried out by Cui, Schwarz, and Datcu ( 2016 ) for multi- temporal SAR image change detection.
Mehr anzeigen

10 Mehr lesen

Combination of LiDAR and SAR data with simulation techniques for image interpretation and change detection in complex urban scenarios

Combination of LiDAR and SAR data with simulation techniques for image interpretation and change detection in complex urban scenarios

In remote sensing, change detection is the process of identifying changes on the earth surface by jointly analyzing two (or more) images of the same geographical area acquired at different times (Bruzzone and Bovolo, 2013). The techniques of change detection have been widely applied in different remote sensing fields, like urban planning, city expansion monitoring, agricultural surveying, natural resource monitoring, natural hazard prevention and damage assessment. Thanks to the remote sensing data acquired by different sensors (optical, LiDAR, SAR) on different platforms (airplanes, satellites), many change-detection techniques have been developed and presented in literature. Some of the techniques have been summarized and classified in Singh (1989), Lu et al. (2004), and Radke et al. (2005). A framework is proposed in Bruzzone and Bovolo (2013) that aims at defining a top-down approach to the design of novel change-detection systems for multi-temporal VHR images. Generally, optical sensors have been extensively exploited for change detection approaches. This is due to the long history of satellite optical sensors since the launch of Landsat in 1972. Besides this, optical data is easier to interpret for human eyes. In contrast, the satellite SAR sensor has only existed since the launch of ERS-1 in 1991. The SAR images are difficult to understand because of their intrinsic speckle phenomenon and geometric distortions.
Mehr anzeigen

126 Mehr lesen

Combination of LiDAR and SAR data with simulation techniques for image interpretation and change detection in complex urban scenarios

Combination of LiDAR and SAR data with simulation techniques for image interpretation and change detection in complex urban scenarios

In remote sensing, change detection is the process of identifying changes on the earth surface by jointly analyzing two (or more) images of the same geographical area acquired at different times (Bruzzone and Bovolo, 2013). The techniques of change detection have been widely applied in different remote sensing fields, like urban planning, city expansion monitoring, agricultural surveying, natural resource monitoring, natural hazard prevention and damage assessment. Thanks to the remote sensing data acquired by different sensors (optical, LiDAR, SAR) on different platforms (airplanes, satellites), many change-detection techniques have been developed and presented in literature. Some of the techniques have been summarized and classified in Singh (1989), Lu et al. (2004), and Radke et al. (2005). A framework is proposed in Bruzzone and Bovolo (2013) that aims at defining a top-down approach to the design of novel change-detection systems for multi-temporal VHR images. Generally, optical sensors have been extensively exploited for change detection approaches. This is due to the long history of satellite optical sensors since the launch of Landsat in 1972. Besides this, optical data is easier to interpret for human eyes. In contrast, the satellite SAR sensor has only existed since the launch of ERS-1 in 1991. The SAR images are difficult to understand because of their intrinsic speckle phenomenon and geometric distortions.
Mehr anzeigen

126 Mehr lesen

Information theoretical similarity measure for change detection

Information theoretical similarity measure for change detection

and a multidimensional density estimation method based on multivariate Edgeworth series expansion are proposed and assessed for the task of multi-temporal change detection. To unify mutual information and variational information, mixed information is proposed to quantify the degree of dependence between two random variables, which are intuitively appropriate for multi-temporal change detection. In the literature, Edgeworth series expansion is widely used in statistics and various engineering fields for one-dimensional density estimation. To compute the mixed information measure, multidimensional density estimation based on multivariate Edgeworth series expansion is proposed and evaluated. Two experiments on real SAR images and optical images are carried out to evaluate the performance of change detection. Experimental results confirm the promising capability of mixed information and the multivariate density estimation based on Edgeworth series expansion.
Mehr anzeigen

4 Mehr lesen

Multi-Temporal Landsat Images and Ancillary Data for Land Use/Cover Change (LULCC) Detection in the Southwest of Burkina Faso, West Africa

Multi-Temporal Landsat Images and Ancillary Data for Land Use/Cover Change (LULCC) Detection in the Southwest of Burkina Faso, West Africa

In order to detect the most adequate period to classify the observed LULC, the year 2011 was selected because it had images from dry season (March), early rainy season (May and June), mid rainy season (July) and late rainy season (October), which cover the entire vegetation period. The results of the mono-temporal RF classifications are shown in Figure 2, and it can be observed that there is an increase in overall, average user’s and average producer’s accuracies from March to October. Table 10 presents the pixels count error matrices obtained from the five mono-temporal classifications. Major confusions were recorded, for instance, between agricultural area and natural vegetation, agricultural area and bare surface and among natural vegetation types. The use of October image for classification outperformed all other mono-temporal attempts, because it reduced confusion between natural vegetation classes (Woodland and Mixed vegetation) and between agricultural area and natural vegetation, among others. The matrices clarify that at the end of the dry period (March) the class agriculture area seems to spectrally resemble other classes, which explains the aforementioned low accuracies obtained for that period. The main finding is that the LULC classification of the late rainy season images (e.g., October) performed better than those applied to images of the mid rainy, early rainy and dry seasons. Therefore, the mono-temporal image of October served as benchmark for assessing the effect of multi-temporal images and ancillary data on LULC classification.
Mehr anzeigen

27 Mehr lesen

Multi-Source and Multi-Temporal Image Fusion on Hypercomplex Bases

Multi-Source and Multi-Temporal Image Fusion on Hypercomplex Bases

on wetlands. Already two years later, the normalized Kennaugh elements were introduced as a consistent frame for the fusion of multi-sensor and multi-polarized SAR data allowing for simple data analysis with standard and open-source software, effective data compression, and even change detection in combination with advanced locally adaptive filtering techniques [ 11 ]. Refining these filtering techniques by multi-directional kernels led to the Schmittlet approach that even enabled the analysis of image texture in the best available (total) intensity image [ 12 ]. The change detection capability was used for mapping wind throw areas in mid-European forests from TerraSAR-X images [ 14 ]. The usefulness of the normalized Kennaugh elements in the context of wetland remote sensing especially in semi-arid regions was proven several times using TerraSAR-X, RADARSAT-2, and even archived data of RADARSAT-1 and ENVISAT-ASAR [ 15 – 17 ]. Due to its flexibility concerning the input data, different polarization combinations could be evaluated at the same time for surface water monitoring purposes [ 18 ]. The advanced preprocessing finally enabled an almost daily estimation of the water gauge in the Forggensee in southern Germany via flooded area [ 19 ]. Furthermore, the Kennaugh decomposition was successfully utilized in coastal applications for the mapping of bivalve beds on exposed intertidal flats from dual-co-polarized TerraSAR-X data [ 20 ]. Going further north, the normalized Kennaugh elements excelled as input feature for classification of Tundra environments in northern Canada. The relation between snow depth, topography, and vegetation was studied on TerraSAR-X time series preprocessed to Kennaugh elements [ 21 ]. Even for the understanding of snow covered sea ice processes from TerraSAR-X the Kennaugh preprocessing showed up to be essential [ 22 ]. Regarding glaciered areas, the polarimetric information captured in Kennaugh elements helped to detect the supraglacial
Mehr anzeigen

37 Mehr lesen

Normalized Compression Distance for SAR Image Change Detection

Normalized Compression Distance for SAR Image Change Detection

With a continuous increase in multi-temporal synthetic aperture radar (SAR) images, leading to enable mapping applications for Earth environmental observation, the number of algorithms for detection of different types of terrain changes has greatly expanded. In this paper, a SAR image change detection method based on normalized compression distance (NCD) is proposed. The procedure mainly consists in dividing two time series images in patches, computing a collection of similarities corresponding to each pair of patches and generating the change map with a histogram-based threshold. The experimental results were computed using 2 Sentinel 1A images over the city of Bucharest, Romania and 2 TerraSAR-X images over the Elbe River and its surrounding area, Germany.
Mehr anzeigen

4 Mehr lesen

A New Coherent Similarity Measure for Temporal Multichannel Scene Characterization

A New Coherent Similarity Measure for Temporal Multichannel Scene Characterization

Abstract—This paper proposes a new method for a measure of coherent similarity between temporal multichannel synthetic aperture radar (SAR) images and its implementation to change detection application. The method is based on mutual infor- mation (MI) from information theory. The MI measures the amount of information in common between coherent temporal multichannel SAR acquisitions. In order to develop an algo- rithm for all kinds of SAR images, such as interferometric SAR, polarimetric–interferometric SAR (PolInSAR), and partial PolIn- SAR, first, the joint density function of temporal multichannel images based on their second-order statistics has been derived. Then, the derived joint density function is used to calculate an analytical expression for the MI between temporal images, which is assumed to be maximal if the temporal images are identical. Although, in this paper, a new coherent similarity measure has analytically been derived for temporal polarimetric SAR images based on complex Wishart process in time, since the mathematical formulation is general, it can equally well be implemented into any kind of multivariate remote sensing data, such as multispec- tral optical and interferometric images after small continuation. This derived quantity has been implemented for change detection application whose aim is to characterize the temporal behavior of the acquisitions. A comparison between the proposed and the other well-known change detection methods by means of scene characterization is shown, describing the advantages due to the fact that the proposed change detector involves almost every facet of applied change detection.
Mehr anzeigen

13 Mehr lesen

Comparative evaluation of signal-based and descriptor-based similarity measures for SAR-optical image matching

Comparative evaluation of signal-based and descriptor-based similarity measures for SAR-optical image matching

which reduces the problem of threshold determination to the question of how to balance the probabilities of: true positives (TP), true negatives (TN), false positives (FP) and false nega- tives (FN). Since TP and TN probabilities are always required to be as high as possible, a trade-off between FP and FN prob- abilities has to be found, which usually depends on the goal one has in mind. In the case of multi-sensor image matching, it is usually much worse to detect a match that is not correct than to miss a correct match, because wrong matches will al- ways negatively affect the final result. Therefore, we seek to minimize the probability for FPs, while we don’t care as much about the FN probability.
Mehr anzeigen

4 Mehr lesen

Field evaluation of a multi-point fibre optic sensor array for methane detection (“OMEGA”)

Field evaluation of a multi-point fibre optic sensor array for methane detection (“OMEGA”)

Figure 8. Response of optical and pellistor head in position 4 to calibration gases: 50% LEL methane followed by air. The optical sensors appeared, in general, to have a slightly faster response than the pellistors, with t 90 less than 10 seconds, which was the time taken for the system processor to take 64 absorption readings and deliver the concentration measurements. The noise level of the optical system also appeared consistently lower than that of the pellistor sensors. However, the calibration span and zero were observed to drift somewhat throughout the program, and were periodically readjusted. Although initially stable, by the end of the test programme the zero response was found to drift significantly over short time periods, the worst example being a drift from zero to +14% LEL over only 30 minutes. Any drift in the response to 50% LEL methane was low in comparison. The zero drift can be explained by higher than expected levels of optical interference in each OMEGA fibre loop. Interference fringes in the spectral scan were interpreted as positive or negative methane peaks, depending on their position, which might vary with the temperature of the interference cavity. Laboratory experiments conducted by GMI confirmed this possibility, showing zero shifts as rapid as the 30-minute change observed in the field.
Mehr anzeigen

9 Mehr lesen

BigEarthNet: A Large-Scale Benchmark Archive for Remote Sensing Image Understanding

BigEarthNet: A Large-Scale Benchmark Archive for Remote Sensing Image Understanding

BigEarthNet is a new large-scale Sentinel-2 benchmark archive, consisting of 590, 326 Sentinel-2 image patches. To construct BigEarthNet, 125 Sentinel-2 tiles acquired between June 2017 and May 2018 over the 10 countries (Austria, Belgium, Finland, Ireland, Kosovo, Lithuania, Luxembourg, Portugal, Serbia, Switzerland) of Europe were initially selected. All the tiles were atmospherically corrected by the Sentinel-2 Level 2A product generation and formatting tool (sen2cor). Then, they were divided into 590, 326 non-overlapping image patches. Each image patch was annotated by the multiple land- cover classes (i.e., multi-labels) that were provided from the CORINE Land Cover database of the year 2018 (CLC 2018).
Mehr anzeigen

4 Mehr lesen

Multi-temporal high-resolution polarimetric L-band SAR observation of a wine-producing landscape

Multi-temporal high-resolution polarimetric L-band SAR observation of a wine-producing landscape

terrain in a wide area of the Frascati wine production. Intensive measurements were carried out on relatively flat parcels imaged at two different incidence angles and belonging to the Pietra Porzia estate [11]. Two areas were selected, one with vine rows almost parallel to the flight ground track, the other nearly orthogonal. Several parameters were measured, including row and vine spacing, leaf height distribution and dimensions, number of leaves per unit area, height distribution and dimension of grapes (Fig.3), number of grapes per unit area, average grape weight, dimensions of trunks and stalks, geometry of supporting poles and wires, roughness and moisture of the terrain, weed height. The selected areas lie within the overlapping region of the two strips, hence are imaged at two different incidence angles, i.e., at about 27° and around 53°.
Mehr anzeigen

3 Mehr lesen

Bigearthnet: A Large-Scale Benchmark Archive for Remote Sensing Image Understanding

Bigearthnet: A Large-Scale Benchmark Archive for Remote Sensing Image Understanding

nity. However, such an approach has several limitations related to the differences on the characteristics of images between computer vision and RS. Additionally, in the existing archives, RS images are annotated by single high-level category labels that are related to the most significant content of the image. However, RS images typically contain multiple classes and thus each image can be simultaneously associated with differ- ent land-cover class labels (i.e., multi-labels). To overcome these problems, we introduce the BigEarthNet that is a new large-scale Sentinel-2 archive 1 and contains 590, 326 Sentinel- 2 image patches. Each patch is annotated with multi-labels provided from the CORINE Land Cover database, which is updated in 2018 (CLC 2018). We propose our archive as a sufficient source for RS image analysis with deep learning. In order to test the BigEarthNet on RS image analysis problems, we focus our attention on image scene classification. To this end, we consider a shallow CNN architecture to be trained on the BigEarthNet. We compare the results obtained by this net- work with the Inception-v2 [1] pre-trained on the ImageNet. We believe that it will make a significant advancement in terms of developments of algorithms for the analysis of large-scale RS image archives.
Mehr anzeigen

5 Mehr lesen

SAR Change Detection in a General Case Using Normalized Compression Distance

SAR Change Detection in a General Case Using Normalized Compression Distance

This approach was used to detect changes in different regions of interest (e.g., the Danube Delta in Romania or Belgica Bank in Greenland) independently of a special scenario or a spec[r]

1 Mehr lesen

Curvelet-based Change Detection for man-made Objects from SAR Images

Curvelet-based Change Detection for man-made Objects from SAR Images

The colour composition in Fig. 3(a) shows three images over the Chuquicamata copper mine in Chile. The regions of high reflectivity caused by an extreme foreshortening effect are slightly smeared over during the geocoding step. One can note that something has changed around the deposit in the middle left of the image. The other images show the results of a curvelet-based change detection with weighted coefficients. In Fig. 3(b) series of red and green stripes along the moun- tain especially in the upper part are visible. In Fig. 3(c) these structures can be found in the lower part. Fig. 3(d), which depicts the changes between the first and the third image can be seen as sum of the two preceding images. The sequence of red (darkened) and green (brightened up) linear features is most remarkable. These changes indicate a systematic dis- placement of linear high backscatterers. An explanation can be found in the geometrical form of the mine: The deposit is built in terraces. If the deposition goes on, the edges of these terraces move and cause also a displacement of the bright lines in the SAR image, because of the reflector-like diplane backscattering at each stage. In contrast to the colour com- position (Fig. 3(a)) the results are very smooth and show no single pixel changes in the surrounding (Fig. 3(b)-3(d)). So, this approach is capable to survey open cast mining activities even though it is not possible at the moment to determine the amount of soil moved.
Mehr anzeigen

4 Mehr lesen

A multiprocessing Framework for SAR Image Processing

A multiprocessing Framework for SAR Image Processing

The framework is used from many different developers from many different countries, most of them barely familiar with C++ programming. The design goal is to provide a simple interface to the numerical operations, e.g. the easies way write the summation of two arrays x and y is:

4 Mehr lesen

Simulation-based Building Change Detection from Multi-Angle SAR Images and Digital Surface Models

Simulation-based Building Change Detection from Multi-Angle SAR Images and Digital Surface Models

Junyi Tao was born in Wuhan, P.R.China, in 1983. He received the Dipl.-Ing.(Univ.) degree in Geodesy and Geoinformatics from the Universit¨at Stuttgart, Germany, in 2009, and the Dr.-Ing. degree in Geodesy and Geoinformatics from Technische Universit¨at M¨unchen, Munich, Germany, in 2015. From October 2009 to 2014, he was a scientific collaborator with the Remote Sensing Technology Institute (IMF), German Aerospace Center (DLR), Oberpfaffenhofen, Germany, working in close co- operation with the chair of Remote Sensing Tech- nology (LMF), Technische Universit¨at M¨unchen (TUM), Munich, Germany. In June 2012, he was a guest scientist at the Remote Sensing Laboratory, Department of Information Engineering and Computer Science, University of Trento, Trento, Italy. He pursued research topics in the field of SAR simulation, SAR image interpretation, multi-modal data fusion, and change detection. In particular, he focused on the combination of LiDAR and SAR data with simulation techniques for object identification and change detection in urban areas. Currently, he is working in the R&D Department of a privately held company, where he is further developing his expertise in sensor data fusion.
Mehr anzeigen

17 Mehr lesen

Interactive clustering for SAR image understanding

Interactive clustering for SAR image understanding

The collected EO data volumes are increasing immensely with a rate of several Terabytes of data a day. With the current EO technologies these figures will be soon amplified, the horizons are beyond Zettabytes of data. The challenge is the exploration of these data and the timely delivery of focused information and knowledge in a simple and understandable format.

1 Mehr lesen

Show all 10000 documents...