The present study focuses on the evaluation of weft knitted fabric roughness using image processing. Various knitted fabric samples have been produced using flat and circular knitting machines. The SMD value of fabrics which refers to surface roughness has been measured by Kawabata surface tester. Images of samples have been taken by a high resolution scanner, converted to grayscale image and then been processed by Matlab; a signal is obtained from each sample. Six different features have been extracted from the signals. The results of correlation test reveal that there is a good correlation between the wave lengths of obtained signal extracted from fabric images and the measured roughness by KES surface tester