Unbalanced JPEG image steganalysis via multiview data match.

•This work focuses on the problem of unbalance JPEG images steganalysis.•A semi-supervised learning algorithm which integrates weighted Fisher linear discriminant and K-means clustering (WFLDK) into a join framework is devised to solve unbalanced image steganalysis.•Multiview match resampling method is proposed to rebalance the unbalanced training images.