Soft morphological filters applied to the removal of noise from CCTV footage

A technique for removing noise from CCTV footage is presented. Soft morphological filters are a type of nonlinear filter. The training of these filters is performed using a Genetic Algorithm. The Genetic Algorithm previously used global measures, for example the Mean Absolute Error and Mean Squared Error to determine the fitness of filters. However, when applied to a variety of data sets such as CCTV footage these measures were found to be inconsistent in producing a near-optimal filter. In this paper the Universal Image Quality Index is used as the fitness measure and has been found to consistently produce near-optimal filters that have improved the quality of the resulting images.