A video directional weighted median filter with adaptive threshold and colour correction

This paper presents a video directional weighted median filter with suitable colour correction to remove random valued impulse noise from colour video sequences. The switching median technique is utilized to protect noise free pixels from filtering so as to avoid blurring of frames. The threshold used for identifying isolated noisy pixels is made adaptive based on the local statistics of the current pixel component. The restoration of noisy pixels is done by utilizing the brightness and colour information obtained from directional weighted median filter and vector directional filter, respectively. Local statistics for video filtering involves a three dimensional sliding window for estimation of characteristics of current pixel. This three dimensional sliding window provides spatial as well as temporal information about the neighbourhood for independent restoration of frame under consideration but decision for filtering is made only on the basis of information obtained from current frame only because of the fact that the consecutive frames may also have special characteristics. Only noise free pixels of three dimensional sliding window, are used for restoration of frame under consideration. Simulation results show that the proposed filter yields superior performance in comparison to other filtering methods available in literature.