Method and apparatus for determination of an approximate standard deviation in an image processor

摘要

An economical and less cumbersome method for the determination of an approximate standard deviation is implemented in hardware that does not require multipliers and root calculators. A mean value for a series of values is first calculated. Then, for each of the values in the series, the difference between the value and the mean value is calculated. Each of the differences is then classified in at least two different groups, with the classification corresponding to the size of the difference. The differences are then weighted according to the group in which they are classified. Finally, the weighted differences are summed and divided by an appropriate predetermined number. This method can be used in systems for noise suppression with a sigma filter.