element mean of matrices

I have 5 matrices of 20*20 size, i want to calculate mean of each element. But i am trying to ommit the least values like <15% from Maximum of each element (i.e if there is any value <15%of maximum on each element of 5 matrices). Plz suggest the process and code

[madhan ravi]

illustrate with a short example

[Leela Sai Krishna]

for example i have a code like following

a=randi(100,1,10); %1*10 matrix with range upto 100

b=rand(100,1,10); %1*10 matrix

c=rand(100,1,10); %1*10 matrix

A=[a;b;c]; % out matrix is 3*10 matrix

A_max=max(A); %output will be 1*10 matrix

i want to apply the condition like if the value on each column shold not be <15% of maximum value on that column.

for eample i have a matrix as follows

A=[100,90,65;

11,68,98;

58,25,8];

i want mean of this after applying the condition(omit <15% of maximum value).

the required output should be like

Out=[79,61,81.5];

Answer 1

[KSSV]

A = rand(5,5,7) ;   % some random data for demo
iwant = zeros(7,1) ;  % initialize the required 
for i = 1:7
    Ai = A(:,:,i) ;   % the present mtrix 
    A_max = max(Ai(:)) ;  % get max of the matrix 
    idx = Ai<15/100*A_max ; % get indices less then 15% of the max 
    iwant(i) = mean(Ai(idx)) ; % get mean 
end

You may do this without defining those many variables. For your understanding I have used the extra variables.

[Leela Sai Krishna]

Thanks for your response

For suppose i have a data like following

val(:,:,1) =
    89
    75
    53
    49
    61
    11
    57
    14
    46
    14
val(:,:,2) =
    11
    84
    99
    11
    83
    10
    31
    46
    45
    34
val(:,:,3) =
    78
    85
    35
    83
    91
    90
    74
    60
    70
    17

then i want mean of 1st element be like (89+78)/2 after omitting <15% of max value. and same for remaining elements also.

If there is no value less than 15% of the maximum then the mean should have to calculate as (N1+N2+N3)/3.

(output value not like (89+11+78)/3 or (89+78)/3) for 1st element. ).

your response is helpful for me, Thanks....