Trimming
From HandWiki
Trimming a data sample consists in removing the n members having the n/2 largest values and the n/2 smallest values of a given parameter. The trimmed mean is the mean value of a parameter for a data sample ignoring the n extreme values. The even positive integer n determines the amount of trimming. When n is one less than the size of the data sample, the trimming is maximum, and the trimmed mean is just the median.
Trimming makes a calculation more robust, i.e. less sensitive to outliers, at the expense of reduced statistical efficiency.
also Winsorization.