Study of ICA algorithm for separation of mixed images
16 May 2012-pp 82-86
TL;DR: In the context of adaptive Neural Network, ICA method tries to train the non-Gaussianity instead of assuming the data to be Gaussian, and this technique is applied to image data and detailed analysis is done for step wise output of the algorithm.
Abstract: The image data can be Gaussian or non-Gaussian or both. If the data is Gaussian then the extraction and processing of image data becomes computationally less complex. Due to this reason many existing techniques like factor analysis, Principle Component analysis, Gabor wavelets etc. assume the data to be Gaussian and processing involves only second order moments such as mean and variance. But if the data is non-Gaussian, then the extraction and processing of image data becomes computationally more complex as it involves higher order moments like kurtosis and a new measure of non-Gaussianity known as negentropy. In this paper a recently developed technique, known as Independent Component Analysis, is applied to image data and detailed analysis is done for step wise output of the algorithm. In the context of adaptive Neural Network, ICA method tries to train the non-Gaussianity instead of assuming the data to be Gaussian.
Citations
More filters
01 Aug 2015
TL;DR: The various challenges involved in exploring the hyperspectral data are described and the seven states of the art methods detailed are discussed.
Abstract: Recently, target detection is a popular research area of hypespectral image processing. An overview of the theory and issues with the analysis of target detection algorithms to exploit hyperspectral imaging data is detailed. First we describe the various challenges involved in exploring the hyperspectral data are discussed. Next we discuss the seven states of the art methods detailed. However target detection has numerous applications. Hence, it is important to pursue research in target detection. In this paper, we review the target detection algorithms in hyperspectral imagery are discussed along with the scope for future research.
13 citations
Cites methods from "Study of ICA algorithm for separati..."
...Using the goodness of kernel methods a kernel based angle-regularized spectral matching algorithm was proposed by [12, 16, 21-22, 29-30]....
[...]
TL;DR: The proposed algorithm more effectively separates change information and reduces computational complexity and the obtained change image has higher accuracy and strong robustness with respect to the background.
Abstract: In order to improve the accuracy and computational efficiency of change detection of multi-temporal remote sensing images, a change detection algorithm based on contourlet transform and independent component analysis (ICA) is proposed. Firstly, multi-scale decomposition of image data is performed by using contourlet transform with multi-scale, directionality and anisotropy. Then independent component analysis is carried out for the decomposed data. And the independent data components are separated by the improved fixed point ICA algorithm based on Newton iteration. Next the separated data components are transformed into image components. Finally, change detection is achieved by threshold segmentation and filtering for change image components. The experimental results show that, compared with the existing three change detection algorithms such as the algorithm based on PCA, the algorithm based on ICA and the algorithm based on wavelet transform and ICA, the proposed algorithm in this paper more effectively separates change information and reduces computational complexity. The obtained change image has higher accuracy and strong robustness with respect to the background.
3 citations
TL;DR: The experimental results show that the proposed algorithm separates change information more effectively and reduces computational complexity, and the obtained change image has higher accuracy and stronger robustness to the background.
Abstract: In order to improve the accuracy and computational efficiency of change detection of multi-temporal remote sensing images, a change detection algorithm based on nonsubsampled contourlet transform (NSCT) and independent component analysis (ICA) is proposed. The flexibility of NSCT in image decomposition and the effectiveness of ICA in image separation are used comprehensively. Firstly, multi-scale decomposition of remote sensing images is performed by NSCT. Then the decomposed low-frequency components and high-frequency components form into partitioned vectors. ICA is carried out for the partitioned vectors and separates mutual independent components. Next the separated components are transformed into image components which include the change image. Finally, change detection result is achieved by threshold segmentation and filtering of the change image. The experimental results show that, compared with the algorithm based on ICA, the algorithm based on wavelet transform and ICA, the proposed algorithm separates change information more effectively and reduces computational complexity. The obtained change image has higher accuracy and stronger robustness to the background.
2 citations
Additional excerpts
...We define the unmixing matrix H, and the output signals HAS HX Y = = are obtained with the strongest independence [10]....
[...]
TL;DR: The problem of separating linearly mixed signals is solved by using filter banks with ICA and it is shown that the proposed algorithm over performs basic ICA.
Abstract: Speech is the fundamental means of communication among humans. Speech production is the process of converting a linguistic message to the acoustic waveform. Separating various linearly mixed speech signals is often modelled by famous cocktail party problem and can be achieved by a technique known as Independent Component Analysis (ICA). ICA is similar to PCA and Factor analysis but it works on non-Gaussian mixture of signals. In this paper, the problem of separating linearly mixed signals is solved by using filter banks with ICA. Comparison of existing ICA technique with the one proposed is done based on experimental results which shows that the proposed algorithm over performs basic ICA.
1 citations
Cites background from "Study of ICA algorithm for separati..."
...[8] Arti Khaparde ,“Study Of ICA Algorithm For Separation Of Mixed Images”, IEEE 2012...
[...]
References
More filters
TL;DR: An efficient algorithm is proposed, which allows the computation of the ICA of a data matrix within a polynomial time and may actually be seen as an extension of the principal component analysis (PCA).
8,522 citations
Additional excerpts
...{ } 2 2 2 2 2 1 1 )}) ( {G E )} ( E{G ( K ) (y) G (E K (y) J U Y − + ≅ (9) Where K1 and K2 are positive constant and U is standardized gaussian variable....
[...]
TL;DR: The basic theory and applications of ICA are presented, and the goal is to find a linear representation of non-Gaussian data so that the components are statistically independent, or as independent as possible.
8,231 citations
TL;DR: Using maximum entropy approximations of differential entropy, a family of new contrast (objective) functions for ICA enable both the estimation of the whole decomposition by minimizing mutual information, and estimation of individual independent components as projection pursuit directions.
Abstract: Independent component analysis (ICA) is a statistical method for transforming an observed multidimensional random vector into components that are statistically as independent from each other as possible. We use a combination of two different approaches for linear ICA: Comon's information theoretic approach and the projection pursuit approach. Using maximum entropy approximations of differential entropy, we introduce a family of new contrast functions for ICA. These contrast functions enable both the estimation of the whole decomposition by minimizing mutual information, and estimation of individual independent components as projection pursuit directions. The statistical properties of the estimators based on such contrast functions are analyzed under the assumption of the linear mixture model, and it is shown how to choose contrast functions that are robust and/or of minimum variance. Finally, we introduce simple fixed-point algorithms for practical optimization of the contrast functions.
6,144 citations
"Study of ICA algorithm for separati..." refers background in this paper
...This implies that { } 2 2 2 1 2 1 E y q q = + = (5) Geometrically, this means that vector q is constrained to be the unit circle on the 2-D plane....
[...]
TL;DR: A novel fast algorithm for independent component analysis is introduced, which can be used for blind source separation and feature extraction, and the convergence speed is shown to be cubic.
Abstract: We introduce a novel fast algorithm for independent component analysis, which can be used for blind source separation and feature extraction. We show how a neural network learning rule can be transformed into a fixedpoint iteration, which provides an algorithm that is very simple, does not depend on any user-defined parameters, and is fast to converge to the most accurate solution allowed by the data. The algorithm finds, one at a time, all nongaussian independent components, regardless of their probability distributions. The computations can be performed in either batch mode or a semiadaptive manner. The convergence of the algorithm is rigorously proved, and the convergence speed is shown to be cubic. Some comparisons to gradient-based algorithms are made, showing that the new algorithm is usually 10 to 100 times faster, sometimes giving the solution in just a few iterations.
3,215 citations
TL;DR: It is shown that a new unsupervised learning algorithm based on information maximization, a nonlinear "infomax" network, when applied to an ensemble of natural scenes produces sets of visual filters that are localized and oriented.
2,354 citations