Fast algorithm to blind and sparkling pixels compensation based on kalman prediction

A blind pixel compensation algorithm based on Kalman Prediction is proposed, and a mathematical model is established. Kalman prediction is used to detect blind pixels in a row data in an image. The blind pixel is displaced by the middle pixel in the neighbor data set. The algorithm is optimized for the real time application. The virtue of the proposed approach is illustrated with a comparison with median filter and mean filter on computation complexity and time consuming. Theory and experiments demonstrate the applicability in real time systems.