We consider here image denoising procedures, based on computationally effective tree-serial param... more We consider here image denoising procedures, based on computationally effective tree-serial parametric dynamic programming procedures, different representations of an image lattice by the set of acyclic graphs and non-convex regularization of a new type which allows to flexibly set a priori preferences. Experimental results in image denoising, as well as comparison with related methods, are provided. A new extended version of multi quadratic dynamic programming procedures for image denoising, proposed here, shows an improved accuracy for images of a different type.
Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170)
Variational Methods in Signal and Image Analysis ... Vadim Mottl, Alexander Blinov, Andrey Kopylo... more Variational Methods in Signal and Image Analysis ... Vadim Mottl, Alexander Blinov, Andrey Kopylov, Alexey Kostin Tula State University ... Ilya Muchnik DIMACS, Center for Discrete Mathematics and Theoretical Computer Science, Rutgers University ... A class of signal and image ...
Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170)
Under the generalized smoothing of a signal or image, a wider class of operations is understood t... more Under the generalized smoothing of a signal or image, a wider class of operations is understood than the plain suppression of noise. A variational approach to the generalized edge-preserving smoothing is considered on the basis of treating the sought-for result of processing as the minimum point of an objective function.
The new approach of relevant feature selection in machine learning is proposed for the case of or... more The new approach of relevant feature selection in machine learning is proposed for the case of ordered features. Feature selection and regularization of decision rule are combined in a single procedure. The selection of features is realized by introducing weight coefficients, characterizing degree of relevance of respective feature. A priori information about feature ordering is taken into account in the form of quadratic penalty or in the form of absolute value penalty on the difference of weight coefficients of neighboring features. Study of a penalty function in the form of absolute value shows computational complexity of such formulation. The effective method of solution is proposed. The brief survey of author's early papers, the mathematical frameworks, and experimental results are provided.
Signal Processing, Pattern Recognition and Applications, 2010
Problems of signal analysis may be practically always considered as those of recovering some hidd... more Problems of signal analysis may be practically always considered as those of recovering some hidden dependences, which are time-varying in the general case. In many situations, the nostationarity mode of the dependence should be expected to change in the observation interval, maybe with some jumps or spikes. In this work, we consider a statistical framework and a family of respective algorithm for timevarying linear regression estimation which preserve essential peculiarities in basically smoothly changing regression coefficients. The method being proposed is simple in tuning and has linear computational complexity with respect to the signal length. In particular, we show how this technique allows to watch the dynamics of the hidden asset composition of an investment portfolio from publicly available data with the purpose of detecting sharp changes in its investment strategy.
Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170), 1998
Variational Methods in Signal and Image Analysis ... Vadim Mottl, Alexander Blinov, Andrey Kopylo... more Variational Methods in Signal and Image Analysis ... Vadim Mottl, Alexander Blinov, Andrey Kopylov, Alexey Kostin Tula State University ... Ilya Muchnik DIMACS, Center for Discrete Mathematics and Theoretical Computer Science, Rutgers University ... A class of signal and image ...
Transactions on Machine Learning and Data Mining, 2009
The new approach of relevant feature selection in machine learning is proposed for the case of or... more The new approach of relevant feature selection in machine learning is proposed for the case of ordered features. Feature selection and regularization of decision rule are combined in a single procedure. The selection of features is realized by introducing weight coefficients, characterizing degree of relevance of respective feature. A priori information about feature ordering is taken into account in the form of quadratic penalty or in the form of absolute value penalty on the difference of weight coefficients of neighboring features. Study of a penalty function in the form of absolute value shows computational complexity of such formulation. The effective method of solution is proposed. The brief survey of author’s early papers, the mathematical frameworks, and experimental results are provided.
Object recognition supported by user interaction for service robots, 2002
Nonlinear transformation of one image plane relative to another by spatially constrained elastic ... more Nonlinear transformation of one image plane relative to another by spatially constrained elastic matching of two pixel grids is proposed a s a t e chnique of measuring image similarity for the purpose of featureless face identi cation. The elastic matching algorithm is devised a s a c ombination of two dynamic programming procedures applied independently to each row and then to each column of the pixel grid. In contrast to the commonly adopted method o f m e asuring face image similarity based on the dynamic link architecture, the proposed method is non-iterative and it avoids image segmentation. Most importantly the method p r ovides the linear computational complexity with respect to the number of pixels without application of parallel computers.
In the classical pattern recognition problem, consideration is given to individual objects, each ... more In the classical pattern recognition problem, consideration is given to individual objects, each of which actually belongs to one of the finite number of classes and is presented for the recognition irrespective of other objects. Recognition objects often form a single interconnected array determined by the nature of the event involved, namely, its natural extent in time or in space along one or a few coordinates. As a consequence, the need arises to take consistent decisions about the classes for all elements of the array. The prior assumption consisting in the fact that neighboring objects more often belong to one class than to different classes will permit one to improve the recognition quality in comparison with the classical case of the independence of classes of separate objects.
We consider here image denoising procedures, based on computationally effective tree-serial param... more We consider here image denoising procedures, based on computationally effective tree-serial parametric dynamic programming procedures, different representations of an image lattice by the set of acyclic graphs and non-convex regularization of a new type which allows to flexibly set a priori preferences. Experimental results in image denoising, as well as comparison with related methods, are provided. A new extended version of multi quadratic dynamic programming procedures for image denoising, proposed here, shows an improved accuracy for images of a different type.
Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170)
Variational Methods in Signal and Image Analysis ... Vadim Mottl, Alexander Blinov, Andrey Kopylo... more Variational Methods in Signal and Image Analysis ... Vadim Mottl, Alexander Blinov, Andrey Kopylov, Alexey Kostin Tula State University ... Ilya Muchnik DIMACS, Center for Discrete Mathematics and Theoretical Computer Science, Rutgers University ... A class of signal and image ...
Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170)
Under the generalized smoothing of a signal or image, a wider class of operations is understood t... more Under the generalized smoothing of a signal or image, a wider class of operations is understood than the plain suppression of noise. A variational approach to the generalized edge-preserving smoothing is considered on the basis of treating the sought-for result of processing as the minimum point of an objective function.
The new approach of relevant feature selection in machine learning is proposed for the case of or... more The new approach of relevant feature selection in machine learning is proposed for the case of ordered features. Feature selection and regularization of decision rule are combined in a single procedure. The selection of features is realized by introducing weight coefficients, characterizing degree of relevance of respective feature. A priori information about feature ordering is taken into account in the form of quadratic penalty or in the form of absolute value penalty on the difference of weight coefficients of neighboring features. Study of a penalty function in the form of absolute value shows computational complexity of such formulation. The effective method of solution is proposed. The brief survey of author's early papers, the mathematical frameworks, and experimental results are provided.
Signal Processing, Pattern Recognition and Applications, 2010
Problems of signal analysis may be practically always considered as those of recovering some hidd... more Problems of signal analysis may be practically always considered as those of recovering some hidden dependences, which are time-varying in the general case. In many situations, the nostationarity mode of the dependence should be expected to change in the observation interval, maybe with some jumps or spikes. In this work, we consider a statistical framework and a family of respective algorithm for timevarying linear regression estimation which preserve essential peculiarities in basically smoothly changing regression coefficients. The method being proposed is simple in tuning and has linear computational complexity with respect to the signal length. In particular, we show how this technique allows to watch the dynamics of the hidden asset composition of an investment portfolio from publicly available data with the purpose of detecting sharp changes in its investment strategy.
Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170), 1998
Variational Methods in Signal and Image Analysis ... Vadim Mottl, Alexander Blinov, Andrey Kopylo... more Variational Methods in Signal and Image Analysis ... Vadim Mottl, Alexander Blinov, Andrey Kopylov, Alexey Kostin Tula State University ... Ilya Muchnik DIMACS, Center for Discrete Mathematics and Theoretical Computer Science, Rutgers University ... A class of signal and image ...
Transactions on Machine Learning and Data Mining, 2009
The new approach of relevant feature selection in machine learning is proposed for the case of or... more The new approach of relevant feature selection in machine learning is proposed for the case of ordered features. Feature selection and regularization of decision rule are combined in a single procedure. The selection of features is realized by introducing weight coefficients, characterizing degree of relevance of respective feature. A priori information about feature ordering is taken into account in the form of quadratic penalty or in the form of absolute value penalty on the difference of weight coefficients of neighboring features. Study of a penalty function in the form of absolute value shows computational complexity of such formulation. The effective method of solution is proposed. The brief survey of author’s early papers, the mathematical frameworks, and experimental results are provided.
Object recognition supported by user interaction for service robots, 2002
Nonlinear transformation of one image plane relative to another by spatially constrained elastic ... more Nonlinear transformation of one image plane relative to another by spatially constrained elastic matching of two pixel grids is proposed a s a t e chnique of measuring image similarity for the purpose of featureless face identi cation. The elastic matching algorithm is devised a s a c ombination of two dynamic programming procedures applied independently to each row and then to each column of the pixel grid. In contrast to the commonly adopted method o f m e asuring face image similarity based on the dynamic link architecture, the proposed method is non-iterative and it avoids image segmentation. Most importantly the method p r ovides the linear computational complexity with respect to the number of pixels without application of parallel computers.
In the classical pattern recognition problem, consideration is given to individual objects, each ... more In the classical pattern recognition problem, consideration is given to individual objects, each of which actually belongs to one of the finite number of classes and is presented for the recognition irrespective of other objects. Recognition objects often form a single interconnected array determined by the nature of the event involved, namely, its natural extent in time or in space along one or a few coordinates. As a consequence, the need arises to take consistent decisions about the classes for all elements of the array. The prior assumption consisting in the fact that neighboring objects more often belong to one class than to different classes will permit one to improve the recognition quality in comparison with the classical case of the independence of classes of separate objects.
Uploads
Papers by A. Kopylov