Wyniki wyszukiwania - BazTech

1

Optimization analysis of rank conditioned rank selection filters

Lukac R.

Journal of Applied Computer Science

|

2004

|

Vol. 12, nr 1

23-42

EN

This paper focuses on she optimization analysis and robustness of a nonlinear filtering class of rank conditioned rank selection (RCRS) filters. which combine the general framework of rank selection filters and rank-order information on the selected input samples. Using the rank selection filter strategy, the output sample is restricted to be an order-statistic from the input set spawned by a sliding filtering window, while the number (known as the order of the filter) and file configuration of selected samples are used to extract she rank-order information lo determine the output ranked sample. By simple varying of the order and configuration of selected samples, the RCRS filler can be designed to perform a number of smoothing operations. As shown in this paper, the order and the configuration of the filter parameters influence the filter robustness, whereas the norm of the optimization criteria affects the RCRS filters in terms of a balance between noise attenuation and detail preserving characteristics.

2

Effective neural LUM smoother for image smoothing applications

Lukac R.

Machine Graphics and Vision

|

2003

|

Vol. 12, No. 3

377-391

EN

In this paper, an effective image filtering approach for the impulsive noise suppression with the simultaneous signal-detail preservation is presented. The novelty of the proposed method lies in the combination of the LUM (lower-upper-middle) smoothing characteristics and the neural network. The included LUM-based impulse detector improves the signal-detail preservation capability of the proposed method, whereas the neural network along with the input LUM smoothers guarantee its noise attenuation capability. Since the LUM operation can be very efficiently implemented, the proposed method is computationally attractive and useful for practical image filtering applications.

3

Switching median filter with a local entropy control

Lukac R., Smolka B., Bieda R., Budzan S.

Journal of Medical Informatics & Technologies

|

2003

|

Vol. 5

MI39--46

EN

This paper presents a new switching median filter utilising local contrast entropy of the samples inside the filtering window. The proposed method is fully adaptive, it requires no optimisation and eliminates the main disadvantages of the local contrast probability based switching median. Excellent performance of the proposed method is a result of the successful analysis of input samples, as the local contrast entropy concept is able to efficiently differentiate between outliers and desired edge samples.

4

Colour video filtering based on vector order-statistics

Lukac R., Smolka B., Bieda R., Budzan S.

Journal of Medical Informatics & Technologies

|

2003

|

Vol. 5

MI31--38

EN

The paper introduces a multichannel filtering approach for colour video denoising taking advantage of an order-statistic theory for vector-valued image signals. The proposed adaptive vector filter utilizes switching between an identity operation and a directional distance smoothing function based on the trimmed set of lowest ranked multichannel samples. Because the proposed method uses the same ordering scheme as the standard vector filters, the computational complexity of the new method is computationally attractive. Moreover, the method outperforms the standard vector filters especially in terms of signal-detail preservation. In this paper, we analyse a three-dimensional filter structure based on a cube filter window spawning 27 samples of three following frames. Thus, the proposed algorithm can be applied for the filtering of spatio-temporal or time-varying vector-valued image signals such as colour image sequences or colour video.

5

Application of the Adaptive Center-Weighted Vector Median Framework for the Enhancement of cdna Microarray Images

Lukac R., Smołka B.

International Journal of Applied Mathematics and Computer Science

|

2003

|

Vol. 13, no 3

369-383

EN

In this paper a novel method of noise reduction in color images is presented. The new technique is capable of attenuating both impulsive and Gaussian noise, while preserving and even enhancing the sharpness of the image edges. Extensive simulations reveal that the new method outperforms significantly the standard techniques widely used in multivariate signal processing. In this work we apply the new noise reduction method for the enhancement of the images of the so called gene chips. We demonstrate that the new technique is capable of reducing the impulsive noise present in microarray images and that it facilitates efficient spot location and the estimation of the gene expression levels due to the smoothing effect and preservation of the spot edges. This paper contains a comparison of the new technique of impulsive noise reduction with the standard procedures used for the processing of vector valued images, as well as examples of the efficiency of the new algorithm when applied to typical microarray images.

6

Optimised directional distance filter

Lukac R.

Machine Graphics and Vision

|

2002

|

Vol. 11, No. 2/3

311-326

EN

In this paper, a new adaptive directional distance filter especially for the impulse noise suppression in color images is provided. The proposed method takes advantage from the optimal filtering situation when only the affected samples are estimated, whereas noisy image points are passed to a filter output without change. For that reason, the proposed method is based on switching between the indentity filter (no smoothing) and the directional distance filter that provides a maximum amount of the smoothing. In order to achieve the most precise control of the proposed method, three center-weighted directional distance filters are utilised to determine a parameter compared with a threshold value. This simple comparison serves as a switching control. After the optimisation of the threshold, it will be shown that the proposed method achieves a significant improvement of the filter performance in comparison with standard vector filter classes for a wide range of impulse noise corruption.

7

Adaptive impulse detection based approaches for the noise reduction in heart image sequences

Smolka B., Lukac R.

Journal of Medical Informatics & Technologies

|

2002

|

Vol. 3

MI21--29

EN

This paper focuses on three-dimensional (3-D) adaptive median filters based on the impulse detection approach designed to effectively remove the impulse noise from cardiographic image sequences. Impulse noise affects the useful information in the form of bit errors and it introduces to the image high frequency changes that prohibit to process and to evaluate the heart dynamics correctly. Therefore biomedical imaging such as vascular imaging and quantification of heart dynamics is closely related to digital filtering. In order to suppress impulse noise effectively, well-known non-linear filters based on the robust order-statistic theory provide interesting results. Although median filters have excellent impulse noise attenuation characteristics, their performance is often accompanied by undesired processing of noise-free samples resulting in edge blurring. The reason is that median filters do not satisfy the superposition property and thus the optimal filtering situation where only noisy samples are affected can never be fully obtained. The presented adaptive impulse detection based median filters, can achieve the excellent balance between the noise suppression and the signal-detail preservation. In this paper, the performance of the proposed approaches is successfully tested for the heart image sequence of 38 frames and the wide range of noise corruption intensity. The results are evaluated in terms of mean absolute error, mean square error and cross correlation.