Adaptive-neighborhood filtering of images corrupted by signal-dependent noise

Abstract

In many image-processing applications the noise that corrupts the images is signal dependent, the most widely encountered types being multiplicative, Poisson, film-grain, and speckle noise. Their common feature is that the power of the noise is related to the brightness of the corrupted pixel. This results in brighter areas appearing to be noisier than darker areas. We propose a new adaptive-neighborhood approach to filtering images corrupted by signal-dependent noise. Instead of using fixed-size, fixed-shape neighborhoods, statistics of the noise and the signal are computed within variable-size, variable-shape neighborhoods that are grown for every pixel to contain only pixels that belong to the same object. Results of adaptive-neighborhood filtering are compared with those given by two local-statistics-based filters (the refined Lee filter and the noise-updating repeated Wiener filter), both in terms of subjective and objective measures. The adaptive-neighborhood approach provides better noise suppression as indicated by lower mean-squared errors as well as better retention of edge sharpness than the other approaches considered.

Intensity-driven adaptive-neighborhood technique for polarimetric and interferometric SAR parameters estimation

“…ANs were also used for gray-level [31], [32] and color images [33], [34] filtering. In the context of SAR imaging, a multidimensional extension has also been proposed for adaptive-neighborhood filtering of multitemporal amplitude data [35] and interferometric coherence and phase images [36], [37].…”

Full Tomographic Reconstruction of 2D Vector Fields using Discrete Integral Data

“…Noise processes of this type are inherent in many fields, such as optics [36], kinematics [37] and magnetic resonance imaging [38]. However, in many cases, for example in telecommunications, the noise that corrupts the data is signal independent.…”

Super‐resolution approach to overcome physical limitations of imaging sensors: An overview

“…However, this assumption is sensitive to abrupt changes of the image intensity where stationarity is not justified. The performance of a nonstationary noise filter in the vicinity of edges depends on how the local statistics are estimated (Jiang and Sawchuk, 1986;Lin et al, 1993;Ozkan et al, 1993;Rangayyan et al, 1998;Song and Pearlman, 1988). The fundamental principle behind the prior local statistics estimators is that the local statistics should be calculated over the neighboring pixels on the same side of the edge over the whole local window.…”

“…After analyzing the system models, the problems can be solved by math-ematical inverse procedure, which is implemented and located right after the output signal nodes as a postprocessor. With this signalprocessing-based approach, image noise can be removed with statistical modeling of the image and the noise (Aiazzi et al, 1998;Jiang and Sawchuk, 1986;Kuan et al, 1985;Lin et al, 1993;Ozkan et al, 1993;Rangayyan et al, 1998;Samy, 1995;Sari-Sarraf and Brzakovic, 1991;Song and Pearlman, 1988); and limited dynamic range can be improved through multiple images of the same scene taken with different exposure times (Bogoni et al, 1999;Debevec and Malik, 1997;Robertson et al, 1999;Yamada et al, 1994). There are signal processing techniques to obtain a high-resolution image from observed multiple low-resolution images; this is called super-resolution image reconstruction (Clark et al, 1985;Eren et al, 1997;Hardie et al, 1997;Hong et al, 1997;Kim and Su, 1993;Komatsu et al, 1993;Park et al, 2003;Patti and Altunbasak, 2001).…”