The membership function can also be represented (based an equation (1)) by the comparison of vector under use with the ideal vector which leads to a distance function is follows.
Where is a distance function.
V. PROPOSED METHODS
With the help of sub filter , the distance function is combined to find fuzzy weights. As these filters are not sensitive with ordinary, this proposed method is more adaptable compared to the existing traditional filters. The sub filter produces output with a vector valued signal which leads to the efficient image restoration
The magnitude distance is represented as
Where defines the angular distance. Here the power parameters P control the importance of the angle criteria versus the distance criteria in the general fuzzy membership function.
We have considered constant P value as 0.23.
VI. RESULTS
Proposed filters are compared with the traditional non linear filters. Noise attenuation properties of various comparable filters are tested with some testing images.
Fig (4) Diatom images With Gaussian. Impulsive and Mixed Gaussian Noise
With mean absolute error, mean square error and the normalized color difference are the majors used in comparison. It is identified that the equal color differences leads to equal distances [7].
Image mixed with Gaussian noise, impulsive noise and mixed Gaussian with impulsive noise are tested with the existing filters [8].counter weighted vector directional sub filters and is defined as
(2)
Here D is a scalar function.
The sub filters plays very vital role to reduce noise in an acceptable fashion.