Image modeling by weighted network can powerfully characterize the structure of images.The node strength of a single node in a weighted network integrates the information on the number and the weights of links incident in it.In order to make good use of structural and statistical information in image analysis,the image is represented by a weighted network.Normally,the dissimilarity between the salt and pepper noise pixel and its nearest neighbor pixels is significant.The noise detection problem has been transformed into a problem of finding the node which has the minimal node strength.An ordered weighted average(OWA) operator is adopted to integrate the information provided by the pixels encompassing the noise pixel for restoration.In this way,a novel algorithm for removing salt and pepper noise in images is developed,where the noise detecting procedure and the filtering procedure are interleaved.Experimental results demonstrate the effectiveness of the proposed algorithm.