Nach oben pdf Comparison of alternative image representations in the context of SAR change detection

Comparison of alternative image representations in the context of SAR change detection

Comparison of alternative image representations in the context of SAR change detection

It is well-known that SAR change detection is not an easy task at all. Due to the geometric and radiometric characteristics of SAR images most of the standard change detection algorithms for optical remote sensing data fail. Some basic SAR change detection techniques, advantages and constraints can be found in [1], which reviews the fundamental approaches. [2] distin- guishes two different types of change detection: amplitude change detection and coherent change detection, exploiting the phase information. The latter type has been examined by [3]. This method presumes a stable phase measurement, so that each incoherent region can be classified as changed. Regarding shorter wave lengths, even a repeat pass acquisi- tion with a very short repetition time (11 days in the case of TerraSAR-X) cannot assure coherence over natural cover. In the case of natural disaster monitoring where reference im- ages often are several years old coherence-based methods are not applicable because too much disturbing incoherence is caused by natural surfaces.
Mehr anzeigen

4 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

In this article, a method based on a non-parametric estimation of the Kullback –Leibler divergence using a local feature space is proposed for synthetic aperture radar (SAR) image change detection. First, local features based on a set of Gabor filters are extracted from both pre- and post-event images. The distribution of these local features from a local neighbourhood is considered as a statistical representation of the local image information. The Kullback –Leibler divergence as a probabilistic distance is used for measuring the similarity of the two distributions. Nevertheless, it is not trivial to estimate the distribution of a high-dimensional random vector, let alone the comparison of two distributions. Thus, a non-parametric method based on k-nearest neighbour search is proposed to compute the Kullback –Leibler diver- gence between the two distributions. Through experiments, this method is compared with other state-of-the-art methods and the e ffectiveness of the proposed method for SAR image change detec- tion is demonstrated.
Mehr anzeigen

10 Mehr lesen

The Schmittlets for automated SAR image enhancement

The Schmittlets for automated SAR image enhancement

The definition of the Schmittlets bases on the hyperbolic tangent function family employed for the normalization of intensity images and image combinations [7], for the description of random distributions in the context of change detection [8], and last but not least for the derivation of filter masks. In contrast to common alternative image representations using gradient operators, Schmittlets exclusively compose out of hyperbolic secant square functions to increase the stability of the analysis. Two different shapes of Schmittlets are predefined: round Schmittlets (equal look number for each direction) and lengthy ones (different look numbers per direction). Round Schmittlets can appear in different scales, i.e. different sizes. Lengthy Schmittlets can additionally adopt different orientations. In accordance with image processing theory the number of distinguishable directions increases from two in the second finest scale to sixteen in the fourth scale [9]. In this manner, 35 Schmittlets are defined from scale zero (original resolution) to scale four (sixteen looks per direction) which are depicted in Fig.1.
Mehr anzeigen

4 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

provided in literature. This dissertation contributes a pixel-based algorithm to detect increased backscattering in SAR images by analyzing the SAR pixel values according to simulated layers. To detect demolished buildings, simulated images are generated using LiDAR data. Two comparison operators (normalized mutual information and joint histogram slope) are used to compare image patches related to same buildings. An experiment using Munich data has shown that both of them provide an overall accuracy of more than 90%. A combination of these two comparison operators using decision trees improves the result. The fourth objective is to detect changes between SAR images acquired with different incidence angles. For this purpose, three algorithms are presented in this dissertation. The first algorithm is a building-level algorithm based on layer fill. Image patches related to the same buildings in the two SAR images are extracted using simulation methods. For each extracted image patch pair, the change ratio based on the fill ratio of building layers is estimated. The change ratio values of all buildings are then classified into two classes using the EM-algorithm. This algorithm works well for buildings with different size and shape in complex urban scenarios. Since the whole building is analyzed as one object, buildings with partly demolished walls may not be detected. Under the same idea, a wall-level change detection algorithm was developed. Image patches related to the same walls in the two SAR images were extracted and converted to have the same geometry. These converted patch pairs are then compared using change ratios based on fill ratio or fill position. Lastly, the wall change results are fused to provide building change result. Compared to the building-level change detection algorithm, this method is more time consuming, but yields better results for partly demolished buildings. A combination of these two algorithms is therefore suggested, whereby the building-level method is used for all buildings and wall-level method additionally for selected large buildings. The third developed algorithm is a wall-level change detection algorithm based on point-feature location. To this end, local maximum points in two SAR images corresponding to the same building façade are compared. This method provides promising result for the present data. It may work better for future data with increased resolution to detect changes of detailed façade structures.
Mehr anzeigen

126 Mehr lesen

Curvelet Approach for SAR Image Denoising, Structure Enhancement, and Change Detection

Curvelet Approach for SAR Image Denoising, Structure Enhancement, and Change Detection

inherent noise is reduced and underlying structures are enhanced depending on their length, their orientation or their intensity. In the image enhancement context this approach is most suitable for fine-structured areas, e.g. city centers. The main problem lies in the determination of thresholds for suppression and em- phasis of structures. The determination of the threshold and the number of coefficients respectively is still experiential and highly dependent on the image content. If the scenes are reconstructed by a fix number of coefficients, the complexity of the scene is restricted. As the image description by the curvelet coefficients is purely based on structures, by omitting coefficients originally smooth areas are often affected by artifacts. At the moment the quadratic weighting of the single curvelet coefficients seems to be the best solution for fully automatic processing chains. The change detection approach provides excellent results in ur- ban areas. The great advantage over pixel based methods is the sensitivity towards changes in structures and the possibility to predefine the scale and the strength of changes to be mapped. Problems occur in natural surroundings like forested areas, where the status of the foliage has an important seasonal impact on the backscattering behavior. Not to mention the weather conditions, snow cover with different moistures can highly modify the ap- pearance in a SAR image. In consequence of that the interpre- tation of the detected changes is very challenging. Although the change images contain clear structures without any disturbances, it is nearly impossible to distinguish man-made from natural, e.g. seasonal, changes, without a priori knowledge about the land cover.
Mehr anzeigen

6 Mehr lesen

Simulation based change detection between DSM and high resolution SAR image

Simulation based change detection between DSM and high resolution SAR image

Because of the slanted acquisition geometry of the SAR sensor, shadowed areas occur in the SAR image very often, especially in images of urban areas. The shadow areas in the simulated image contain no signal, i.e. the grey value for shadow is zero. For the same area in the SAR image, the intensity should be very low, in comparison to other pixels. If some new buildings are built in or near the “shadow area” between the acquisition time of the DSM and SAR, there will be high intensity of pixels in the “shadow area”. Through detection of these pixels we can find the positive changes.
Mehr anzeigen

4 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

provided in literature. This dissertation contributes a pixel-based algorithm to detect increased backscattering in SAR images by analyzing the SAR pixel values according to simulated layers. To detect demolished buildings, simulated images are generated using LiDAR data. Two comparison operators (normalized mutual information and joint histogram slope) are used to compare image patches related to same buildings. An experiment using Munich data has shown that both of them provide an overall accuracy of more than 90%. A combination of these two comparison operators using decision trees improves the result. The fourth objective is to detect changes between SAR images acquired with different incidence angles. For this purpose, three algorithms are presented in this dissertation. The first algorithm is a building-level algorithm based on layer fill. Image patches related to the same buildings in the two SAR images are extracted using simulation methods. For each extracted image patch pair, the change ratio based on the fill ratio of building layers is estimated. The change ratio values of all buildings are then classified into two classes using the EM-algorithm. This algorithm works well for buildings with different size and shape in complex urban scenarios. Since the whole building is analyzed as one object, buildings with partly demolished walls may not be detected. Under the same idea, a wall-level change detection algorithm was developed. Image patches related to the same walls in the two SAR images were extracted and converted to have the same geometry. These converted patch pairs are then compared using change ratios based on fill ratio or fill position. Lastly, the wall change results are fused to provide building change result. Compared to the building-level change detection algorithm, this method is more time consuming, but yields better results for partly demolished buildings. A combination of these two algorithms is therefore suggested, whereby the building-level method is used for all buildings and wall-level method additionally for selected large buildings. The third developed algorithm is a wall-level change detection algorithm based on point-feature location. To this end, local maximum points in two SAR images corresponding to the same building façade are compared. This method provides promising result for the present data. It may work better for future data with increased resolution to detect changes of detailed façade structures.
Mehr anzeigen

126 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 recent SAR systems like TerraSAR-X do not only allow a weather and illumination independent availability but also the capture of large coverages with high resolution. Hence, SAR sensors excel in the survey of changes in vulnerable struc- tures, e.g. urban infrastructure. In the case of natural disas- ters a fast data acquisition becomes possible, but in order to provide a final product, i.e. damage map or accessibility map for roads, also ”near realtime” a fully automatic data process- ing and interpretation is needed. A main problem for SAR image interpretation apart from the geometrical aspect is the high noise level caused by the speckle effect inter alia. The reduction of noise, e.g. by the multi-looking approach, often goes along with a loss of resolution. While structure preserv- ing filters do not enhance fine structured areas, smoothening filters even blur the structures apparent in SAR data over ur- ban areas. So, resolution and structure preserving filter al- gorithms are still a research topic. In this context alternative
Mehr anzeigen

4 Mehr lesen

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

Wetland Monitoring Using the Curvelet-Based Change Detection Method on Polarimetric SAR Imagery

Wetland Monitoring Using the Curvelet-Based Change Detection Method on Polarimetric SAR Imagery

Abstract: One fundamental task in wetland monitoring is the regular mapping of (temporarily) flooded areas especially beneath vegetation. Due to the independence of weather and illumination conditions, Synthetic Aperture Radar (SAR) sensors could provide a suitable data base. Using polarimetric modes enables the identification of flooded vegetation by means of the typical double-bounce scattering. In this paper three decomposition techniques—Cloude-Pottier, Freeman-Durden, and Normalized Kennaugh elements—are compared to each other in terms of identifying the flooding extent as well as its temporal change. The image comparison along the time series is performed with the help of the Curvelet-based Change Detection Method. The results indicate that the decomposition algorithm has a strong impact on the robustness and reliability of the change detection. The Normalized Kennaugh elements turn out to be the optimal representation for Curvelet-based change detection processing. Furthermore, the co-polarized channels (same transmit and receive polarization in horizontal (HH) and vertical (VV) direction respectively) appear to be sufficient for wetland monitoring so that dual-co-polarized imaging modes could be an alternative to conventional quad-polarized acquisitions.
Mehr anzeigen

16 Mehr lesen

Interannual change detection of Mediterranean seagrasses using RapidEye image time series

Interannual change detection of Mediterranean seagrasses using RapidEye image time series

remotely sensed data. The remote sensing of seagrasses lying in optically shallow waters (where the observed surface reflectance contains signal from the bottom in contrast to an optically deep column) faces a plethora of inherent obstacles due to the complex nature of the media above the seagrass beds themselves. Obstacles like water column constituents, sunglint, and skyglint presence, air-water interface interference could impede the detection of seagrasses and require, usually, consideration through relevant algorithms. The presented methodological workflow could act like an alternative ecological assessment showing current trends, revealing regressing seagrasses, and allowing better conservation of these complex but also significant ecosystems. Potential improvements in the given approach could be the existence of in situ optical measurements of several relevant parameters, broader bathymetry field data, advanced radiative transfer simulations, possible comparison of different machine learning algorithms for the improvement of classification and identification of seagrasses and better tuning of those algorithms. Currently, seagrasses are decreasing in alarming rates in a global scale. Linkage of this decreasing trend with the anthropogenic and natural interference through Earth observation of climate change, eutrophication, coastal development as well as temperature, salinity, and hydrodynamic change could develop and refine machine learning models to
Mehr anzeigen

15 Mehr lesen

Shape representations for image based applications

Shape representations for image based applications

t(I 0 (p  0 )) ≈ t(I 1 (p  1 )) (2.4) where t is suitable transfer function depending on the particular illumination and surface model. For instance, assuming a constant brightness of a surface point in all images, i.e., the above mentioned Lambertian surface model and identical internal camera parameters like exposure times, t is simply the identity function [KS00b]. To address deviations from the constant brightness assumption the transfer function t could be a conversion to a different color space or the normalization of color values within small patches around the image positions p  i [Her04]. More sophisticated functions t can be employed when more information about the photometric model of the scene is available (e.g., [YPW03, HS05]). There are a variety of standard techniques for detecting explicit image correspon- dences, ranging from single feature detection and tracking [BM04] over dense optical flow [BBPW04] to complex scale and illumination invariant feature descriptors [BTV06, Low04]. However, the following chapters will introduce a number of new, robust and il- lumination invariant transfer and comparison functions for correspondence evaluation to address the specific reconstruction problems discussed in this thesis. Since these corre- spondence measures evaluate the consistency of the appearance of a point p in different images, it will also be referred to as the photo-consistency or surface confidence φ(p) in the following chapters.
Mehr anzeigen

179 Mehr lesen

Image Content Dependent Compression of Polarimetric SAR Data

Image Content Dependent Compression of Polarimetric SAR Data

The algorithm uses wavelet transform to efficiently code the wavelet coefficients where in contrary to many imple- mentation also the coarse scale image is lossy coded. This allows - for a given bit rate – to allocate more bits to the higher subbands. This is important to preserve the details and the phase information, because important information is also represented in the high frequency subbands if wavelet trans- form is applied on real and imaginary part of single-look complex SAR data.

3 Mehr lesen

Change detection in streaming data analytics: a comparison of Bayesian online and martingale approaches

Change detection in streaming data analytics: a comparison of Bayesian online and martingale approaches

of the initial threshold. Alternatively, different martingale tests with different thresholds and composite change detection criteria can be employed (Ho and Wechsler, 2007b). However, while this may improve performance, it also increases the risk of adding false alarms produced by individual martingale tests. A modified martingale algorithm (Table 1) is introduced in this paper to address some of the challenges facing the original approach. The motivation for the modification is to address a common challenge for martingale approaches, related to the time delay between the true change point and the change point detected by the test. The potential significance of this often depends on the application context. In long run processes, the estimation of the strangeness becomes computationally expensive. This is due to the fact that when the frequency of changes is low or when there are no changes, the size of the buffer set of samples (T in Table 1), used to compute the strangeness, can become excessively large. To improve the computational efficiency of estimating the strangeness, down- sampling (Chawla, 2009) and windowing techniques can be used. Engineering processes often exhibit rapid changes in a limited number of time steps before going back to a steady- state. For example, a driver’s acceleration is a typical case. A driving journey may switch between coasting and cruising to modes that involve gear change, acceleration and deceleration with braking. In such cases, the growth of the martingale value is not fast enough to capture the change. Lowering the detection threshold will result in confusing true changes with noise, resulting in higher false detections.
Mehr anzeigen

6 Mehr lesen

Measuring the Performance of Corporate Acquisitions: An Empirical Comparison of
Alternative Metrics

Measuring the Performance of Corporate Acquisitions: An Empirical Comparison of Alternative Metrics

correlations suggest this association does not hold in practice. Indeed, inspection of the data relating to divestments within the first six years, i.e. broadly coincident to the time period in which the managers’ assessments were collected, revealed that 29% of these divestments related to acquisitions that were viewed as very successful by the managers themselves (>=4.0 on the Likert scale). Overall, these results support the view that rather than failure, divestment in some

22 Mehr lesen

Detection of traffic congestion in airborne SAR imagery

Detection of traffic congestion in airborne SAR imagery

Remote sensing sensors installed on aircrafts or satellites enable information collection on a large scale and thus seem to be very suitable for various traffic monitoring applications. Optical systems are already in use, e.g. [2], but are quite limited due to their daylight operation and cloud-free conditions requirements and the need of operator interaction during the data processing. Synthetic aperture radar (SAR) sensors due to their all-weather capabilities seem to be well suited for such type of applications. Ground moving target indication (GMTI) approaches based on the DPCA technique are currently
Mehr anzeigen

5 Mehr lesen

Romanian wine sector in the context of climate change

Romanian wine sector in the context of climate change

At the beginning of forced ripening red varieties are manifested by the grains before they reach the size of typical of the variety. If hydric deficit in the short term, it slows the restore and colorize the beans continues increasing. Otherwise, the grapes, the leaves are fading partially or totally dry and fall one by one, the grape production is thus make Spiramycin less dramatically. The present severe burns with foliar appearance of dryness, which affects the major process of photosynthesis and accumulation of the active substances that encourage the maturation of the grapes.
Mehr anzeigen

5 Mehr lesen

Food security in the context of climate change in Pakistan

Food security in the context of climate change in Pakistan

2. Review of Literature Different studies conducted in the world show significant impact of climate change on agriculture in general and wheat production in particular. Regions with massive grain production may have a negative effect due to increase in temperature. Further, crop production relies on the inter-regional adjustments of the region (Tobey et al., 1992). Wheeler & von Braun (2013) argue that short-term variability in food supply may have adverse effects on sustainability of food systems in the world under climate change scenario. Climate change can have devastating effect on already vulnerable and malnourished regions of the world. Qureshi et al. (2013) find negative impact of climate change on Australian food exports, indicating that global food security would be affected because of sustainable Australia’s contribution to international trade in wheat, meat and dairy products. Hanjra & Qureshi (2010) examine linkages food security and global water supply and show that water availability for food security would decline to an alarming situation in the absence of right action. Investment in research, infrastructure development and conservation practices is suggested to reduce intensity of water shortage in the coming years.
Mehr anzeigen

12 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

Single and multiple change point detection in spike trains : comparison of different CUSUM methods

Single and multiple change point detection in spike trains : comparison of different CUSUM methods

FIGURE 4 | Performance of all six CUSUM versions and the Rate Change method. (A–D) Single stimulus change detection with the optimized parameters listed in Table 4A. (E–H) Multiple stimulus change detection with the optimized parameters listed in Table 4B. Boxplots (A–C,E–G) show distribution of performances obtained for the 10 iterations of the stimulus protocol. For each stimulus trial, the performance was calculated by the percentages of correct and false events for all 728 stimulus changes (A,B,E,F) or all stimulus changes included in each of the clusters, respectively (C,G). Boxplots (D,H) display distributions of all individual detection times obtained in all 10 stimulus iterations (see legend of Figure 3 for number of stimulus changes). (A) Boxplots of the total performances P of Equation (7); (B) Relative frequencies of the correct (E true ), missed (E no ), false (E false ), too early (E early ) and too late (E late ) events. E true and E false were calculated from Equation (6) and E no , E early and E late were determined as described in Section 3.2.1. Note that E true + E no + E false = 1 and E early + E late = E false . (C) Boxplots of the correct events for the different response clusters. The mean cluster PSTHs are illustrated in the top. (D) Boxplots of the detection time of the change points for the correct events relative to the stimulus change for each cluster. Note that the allowed range of detection times was between -5 and 90 ms because of the latency adjustment (see Sections 2.1 and 2.3). (E) Boxplots of the total performances P of Equation (7) for multiple stimulus change detection, (F) Relative frequencies of the correct E true , missed (E missed ), false (E false ), double (E double ) and stochastic (E stoch ) events. E true and E false were calculated from Equation (6) and E missed , E double and
Mehr anzeigen

18 Mehr lesen

Show all 10000 documents...