Weighted vector median operation for filtering multispectral data

Weighted vector median and weighted extensive vector median operations for filtering multispectral data are proposed. The operations share the good qualities of vector median filters, but allow extra freedom in the filter design. The weights of the input samples are defined as vectors. With this structure the possible difference in the noise statistics in the different signal components can be taken into account. An example of the weighted vector median for color image filtering is given.<>