how to filter quantization uncertainty

what is the best option to filter data with quantization uncertainty.if the curve bounces between two values before an edge. what I want is to have the mean of upper and lower value, which is not possible with med1 median filter, because the median of e.g. 99*upper value + 100*lower value is lower value and not the mean. what filter returns the mean while preserving sharp edges?

thanks for your help!

Answer 1

[Star Strider]

Assuming I understand your Question, I would use a moving average filter. Experiment with the length to get the result you want:

s = [5 + randi([0 1], 1, 20)  4:-1:0];                      % Create Data
t = 0:length(s)-1;

N = 4;                                                      % Filter Length
b = ones(1,N);
a = N;
sf = filtfilt(b, a, s);

figure(1)
plot(t, s)
hold on
plot(t, sf)
hold off
grid
axis([xlim    0  10])