FEFROST -- Enhanced Frost SAR Speckle Filter

See Also: FSPEC, FAV, FED, FGA, FME, FMO, FSHARP, FSOBEL, FPR, FPRE

INPUT PORT(S)

Input

Contains the input image layer(s) consisting of the data to be filtered.

Mask

Contains the input area mask layer, that is, the bitmap that indicates which area in the input raster should be processed. If no bitmap is provided the entire layer is processed.

OUTPUT PORT(S)

Output

Contains the output layer(s), that will received the filtered image(s).

INPUT PARAMETER(S)

Image Type

Specifies the image format of the radar image, which can be supplied in either the amplitude or power (intensity) format.

Amplitude is the square root of power. Most radar images are supplied in the amplitude format to preserve the values.

Filter X Size (Pixels)

Specifies the filter's x size. The valid values must be odd integers.

Filter Y Size (Pixels)

Specifies the filter's y size. The valid values must be odd integers.

Number of Looks

Specifies the (effective) number of looks of the image. This is used to derive noise variance. By adjusting NumberLooks, the user can control the amount of smoothing applied to the image.

Damping Factor

Specifies the damping constant for the filter. This constant specifies the extent of the damping effect of the filtering. The default value of 1 is sufficient for most SAR images.

The use of large values for DampFactor allows for better preservation of sharp edges, but reduces the smoothing effect. The use of small values for DampFactor increases the smoothing effect, but does not preserve sharp edges well.

DETAILS

The filter size can be specified through the FilterSize parameters. Different filter sizes will greatly affect the quality of processed images. If the filter is too small, the noise filtering algorithm is not effective. If the filter is too large, subtle details of the image will be lost in the filtering process. A 7x7 filter usually gives the best results.

The NumberLooks parameter is used to estimate noise variance and it effectively controls the amount of smoothing applied to the image by the filter. Theoretically, the correct value for NumberLooks should be the effective number of looks of the radar image. It should be close to the actual number of looks, but may be different if the image has undergone resampling. The user may experimentally adjust the NumberLooks value so as to control the effect of the filter. A smaller NumberLooks value leads to more smoothing; a larger NumberLooks value preserves more image features.

A damping factor (DAMP) is required by the Enhanced Frost filter. The value of DAMP defines the extent of exponential damping (the smaller the value, the smaller the damping effect). It depends on the non-filtered image and may require trial-and-error experiments to determine the best value. The default value for DAMP is 1.

FEFROST performs spatial filtering on each individual pixel in an image using the grey level values in a square window surrounding each pixel. The dimensions of the filter must be odd, and must be at least 3x3.

All pixels are filtered. In order to filter pixels located near the edges of the image, edge-pixel values are replicated to give sufficient data.

A bitmap specifies the area within the input layer which will be processed. Only this area will be filtered and the rest of the image will be unchanged. If no bitmap is connected, the entire database is processed.

The Enhanced Frost filter model requires that the signal represents power. If the input image is in amplitude format, each grey level will be squared to derive power and finally square root will be applied to the filtered result.

ALGORITHM

 Jong-Sen Lee, "Digital Image Enhancement and Noise Filtering
 by Use of Local Statistics", IEEE Transactions on Pattern
 Analysis and Machine Intelligence, Vol. PAM1-2, No. 2, March, 1980.

 J.S.Lee, "Refined Filtering of Image Noise Using Local Statistics"
 Computer Graphic and Image Processing 15, 380-389 (1981)

 D.T. Kuan, A.A. Sawchuk, T.C. Strand, and P. Chavel, 
 "Adaptive restoration of images with speckle," IEEE Trans. ASSP.,
 Vol. 35, no. 3, pp. 373-383, March 1987.

 A. Lopes, R. Touzi and E. Nezry, "Adaptive speckle filters and
 Scene heterogeneity", IEEE Transaction on Geoscience and Remote
 Sensing, Vol. 28, No. 6, pp. 992-1000, Nov. 1990.

 V.S. Frost, J.A. Stiles, K.S. Shanmugan, and J.C. Holtzman,
 "A model for radar images and its application to adaptive digital 
 filtering of multiplicative noise," IEEE Trans. Pattern Analysis
 and Machine Intelligence, vol. 4, no. 2, pp. 157-166, March 1982. 

 A. Lopes, E. Nezry, R. Touzi, and H. Laur, "Structure detection
 and statistical adaptive speckle filtering in SAR images",
 International Journal of Remote Sensing, Vol. 14, No. 9, 
 pp. 1735-1758, 1993.

 A. Lopes, R. Touzi and E. Nezry, "Adaptive speckle filters and
 Scene heterogeneity", IEEE Transaction on Geoscience and Remote
 Sensing, Vol. 28, No. 6, pp. 992-1000, Nov. 1990.

 Zhenghao Shi and Ko B. Fung, 1994, "A Comparison of Digital
 Speckle Filters", Proceedings of IGRASS 94, August 8-12, 1994.

ENHANCED FROST FILTER

The filter output is:

       R = Im                  for Ci <= Cu
       R = Rf                  for Cu < Ci < Cmax
       R = Ic                  for Ci >= Cmax

       M = exp(- A * T)

       A   =   DAMP * (Ci-Cu)/(Cmax-Ci)
       Ci  =   S / Im
       Cu  =   SQRT(1/NLOOK)
       Cmax =  SQRT(1+2/NLOOK)         
       Im  =   mean value of intensity within window
       S   =   standard deviation of intensity within window
       T   =   the absolute value of the pixel distance
               from the centre pixel to its neighbours in 
               the filter window       

       DAMP =  exponential damping factor

       Rf = (P1*M1 + P2*M2 + ... +  Pn*Mn) / (M1 + M2 + ... + Mn)

       P1 .. Pn   are grey levels of each pixel in filter window

       M1 .. Mn   are weights (as defined above) for each pixel