Improved switching median filter for removing impulse noise from the image of the welded joint

To suppress the impulse noise in the image of the welded joint effectively, the improved switching median (ISM) filter is proposed. In the algorithm, the local statistics based noise detector is adopted to classify the pixels in the image into uncorrupted pixels and noise pixels. The noise-free pixels are left unprocessed. The number of noise pixels is counted to estimate the noise density in the image, based on which the suitable filtering means is adopted to remove the impulse noise. Experimental results show that the ISM filter can suppress noise effectively while preserving fine details very well. The ISM filter gains the obvious advantage over the switching median (SM) filter, filter based on extremum and median value (EM) and progressive switching median(PSM) filter, especially when the density noise in the image is relatively high.