Piecewise Linear Model-Based Image Enhancement
- PDF / 4,436,604 Bytes
- 9 Pages / 600 x 792 pts Page_size
- 49 Downloads / 188 Views
Piecewise Linear Model-Based Image Enhancement Fabrizio Russo Department of Electrical, Electronic and Computer Engineering (DEEI), University of Trieste, Via Valerio 10, Trieste 34127, Italy Email: [email protected] Received 1 September 2003; Revised 23 March 2004 A novel technique for the sharpening of noisy images is presented. The proposed enhancement system adopts a simple piecewise linear (PWL) function in order to sharpen the image edges and to reduce the noise. Such effects can easily be controlled by varying two parameters only. The noise sensitivity of the operator is further decreased by means of an additional filtering step, which resorts to a nonlinear model too. Results of computer simulations show that the proposed sharpening system is simple and effective. The application of the method to contrast enhancement of color images is also discussed. Keywords and phrases: image enhancement, sharpening, noise reduction, nonlinear filters.
1.
INTRODUCTION
It is known that a critical issue in the enhancement of images is the noise increase that is typically produced by the sharpening process [1]. A classical example is represented by the linear unsharp masking (UM) method. Since a fraction of the high-pass filtered image is added to the original data, the resulting effect produces edge enhancement and noise amplification as well. In order to address this issue, more effective approaches resort to nonlinear filtering that can realize a better compromise between image sharpening and noise attenuation [2, 3, 4, 5, 6]. In particular, weighted medians (WMs) have been successfully experimented as a replacement for high-pass linear filters in the UM scheme [7]. In this framework, methods based on permutation weighted medians (PWMs) offer very interesting results because they can prevent the noise amplification during the enhancement process [8, 9]. Polynomial UM approaches constitute another family of nonlinear methods for image enhancement. Interesting examples include the Teager-based operator [10, 11] and the cubic UM technique [12]. Rational UM [13] represents a powerful approach to contrast enhancement. It can avoid noise amplification and excessive overshoot on sharp details. Nonlinear methods based on fuzzy models have also been investigated. Indeed, fuzzy systems are well suited to model the uncertainty that occurs when conflicting operations should be performed, for example, detail sharpening and noise cancellation [14, 15, 16]. The most effective approaches can enhance the image data without increasing the noise. However, their ability to reduce the noise during the sharpening process is limited. In this respect, methods based
on forward and backward (FAB) anisotropic diffusion constitute a powerful class of enhancement techniques [17, 18]. Since anisotropic diffusion is typically an iterative process, the noise can be progressively reduced by means of an appropriate choice of parameter settings. In this paper, a new simple technique for the enhancement of noisy images is presented. The proposed method impr
Data Loading...
