Maximum Homogeneity over a Pixel Neighborhood for Image Filtering

Image filtering using switched mean filter with most homogeneous gray values in neighborhood.

fcnMaxHomogeneityOverPixelNeighborhood performs noise filtering on an image using a switched mean filter with most homogeneous gray-values within a pre-specified local neighborhood around the pixel of interest. It uses a square neighborhood of 3x3 pixels to estimate the gray-level statistics in default settings.

Method is based on

[1] S. Ali and R. Burge, "New automatic techniques for smoothing and segmenting sar images," Signal Processing, vol. 14, no. 4, pp. 335–346, 1988. [http://dx.doi.org/10.1016/0165-1684(88)90092-8]

Implementation is based on

[1] D. Sheet, S. Pal, A. Chakraborty, J. Chatterjee, A.K. Ray, "Image quality assessment for performance evaluation of despeckle filters in Optical Coherence Tomography of human skin," 2010 IEEE EMBS Conf. Biomedical Engineering and Sciences (IECBES), pp.499-504, Nov. 30 2010 - Dec. 2 2010. [http://dx.doi.org/10.1109/IECBES.2010.5742289]

[2] D. Sheet, S. Pal, A. Chakraborty, J. Chatterjee, A.K. Ray, "Visual importance pooling for image quality assessment of despeckle filters in Optical Coherence Tomography," 2010 Intl. Conf. Systems in Medicine and Biology (ICSMB), pp.102-107, 16-18 Dec. 2010. [http://dx.doi.org/10.1109/ICSMB.2010.5735353]