Simulation framework for spatio-spectral anomalous change detection

Theiler, James; Harvey, Neal R.; Porter, Reid B.; Wohlberg, Brendt

doi:10.1117/12.818632

Cited by 18 publications

(28 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is a useful paradigm, and that formalism has been extended to the problem of anomalous change detection [4], [11]. Write 3 " 9 @ 4 6 5 ¢ 7…”

Section: B Anomaly Detection As Binary Classificationmentioning

confidence: 99%

“…[2], [3] have argued that the interesting changes are the anomalous changes, and Ref. [4] proposed a framework that built on the machine learning formalism for anomaly detection, but recast the problem in terms of binary classification: pervasive differences versus anomalous changes. This paper will take that same point of view, but will consider the more extreme case that the anomalous changes are smaller than a pixel.…”

Section: Introductionmentioning

confidence: 99%

“…[4]. But by taking smaller values of ¢ , the anomalous change detector can be optimized for subpixel anomalies.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Subpixel Anomalous Change Detection in Remote Sensing Imagery

Theiler

2008

2008 IEEE Southwest Symposium on Image Analysis and Interpretation

Self Cite

View full text Add to dashboard Cite

Abstract-A machine-learning framework for anomalous change detection is extended to the situation in which the anomalous change is smaller than a pixel. Although the existing framework can be applied to (and does have power against) the subpixel case, it is possible to optimize that framework for the subpixel case when the size of the anomalous change is known. The limit of infintesimally small anomaly turns out to be welldefined, and provides a new parameter-free anomalous change detector which is effective over a range of subpixel anomalies, and continues to have reasonable power against the full-pixel case.

show abstract

“…It is a useful paradigm, and that formalism has been extended to the problem of anomalous change detection [4], [11]. Write 3 " 9 @ 4 6 5 ¢ 7…”

Section: B Anomaly Detection As Binary Classificationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Subpixel Anomalous Change Detection in Remote Sensing Imagery

Theiler

2008

2008 IEEE Southwest Symposium on Image Analysis and Interpretation

Self Cite

View full text Add to dashboard Cite

show abstract

“…A recently developed machine learning framework [7] extended this change detection methodology to arbitrary data distributions, and even for Gaussian distributions was shown to exhibit improved performance [8]. But the ACD algorithms in Refs.…”

Section: Introductionmentioning

confidence: 99%

Detection of ephemeral changes in sequences of images

Theiler

Adler‐Golden

2008

2008 37th IEEE Applied Imagery Pattern Recognition Workshop

Self Cite

View full text Add to dashboard Cite

Abstract-The formalism of anomalous change detection, which was developed for finding unusual changes in pairs of images, is extended to sequences of more than two images. Extended algorithms based on RX, Chronochrome, and Hyper are presented for identifying the most anomalously changing pixels in a sequence of co-registered images. Experimental comparisons are performed both on real data with real anomalies and on real data with simulated anomalies.

show abstract

“…The more interesting changes, on the other hand, are often anomalous and involve only a few pixels in the image. Schaum and Stocker (1997) and Clifton (2003) have argued that the interesting changes are the anomalous changes and (Theiler and Perkins, 2006;Hwang et al, 2008) proposed a framework that built on the machine learning formalism for anomaly detection, but recast the problem in terms of binary classification: pervasive differences versus anomalous changes.…”

Section: Introductionmentioning

confidence: 99%

Neural Network Change Detection Model for Satellite Images Using Textural and Spectral Characteristics

Helmy¹,

El-Taweel²

2010

American J. of Engineering and Applied Sciences

View full text Add to dashboard Cite

Problem statement:Change detection is the process of identifying difference of the state of an object or phenomena by observing it at different time. Essentially, it involves the ability to quantify temporal effects using multi-temporal data sets. Information about change is necessary for evaluating land cover and the management of natural resources. Approach: A neural network model based on both spectral and textural analysis is developed. Change detection system in this study is presented using modified version of back-propagation-training algorithm with dynamic learning rate and momentum. Through proposed model, the two images at different dates are fed into the input layer of neural network, in addition with Variance, Skewness and Eculedian for each image that represent different texture measure. This leads to better discrimination process. Results: The results showed that the trained network with texture measures achieve 23% higher accuracy than that without textural parameters. Conclusion: Adding textural parameters of satellite images through training phase increases the efficiently of change detection process also, it provides adequate information about the type of changes. It also found, when using dynamic momentum and learning rate, time and effort needed to select their appropriate value is reduced.

show abstract

Simulation framework for spatio-spectral anomalous change detection

Cited by 18 publications

References 0 publications

Subpixel Anomalous Change Detection in Remote Sensing Imagery

Subpixel Anomalous Change Detection in Remote Sensing Imagery

Detection of ephemeral changes in sequences of images

Neural Network Change Detection Model for Satellite Images Using Textural and Spectral Characteristics

Contact Info

Product

Resources

About