An implementation of the principal components transformation is presented which makes use of analogue and digital electronic hardware under the control of a microprocessor. The hardware is used to collect covariance statistics from a multichannel set of digital video images and to perform the actual transformation on real-time analogue video signals derived from the stored data. This approach offers considerable advantages over a completely software based system in terms of speed and cost. An example of the use of the system for the analysis of Landsat multispectral data is presented.