In　order　to　embed　data　into　binary　images,　some　pixels　need　to　be　modified.　Modification　of　these　pixels　will　cause　visual　distortion.　This　paper　proposes　a　novel　approach　to　measure　the　distortions　introduced　by　modifying　individual　pixels　in　binary　images　and　the　approach　can　be　used　as　part　of　a　watermark　embedding　process.　The　distortion　scores　of　pixels　are　determined　by　the　occurrence　frequencies　of　3x3　patterns　in　original　binary　image.　Experiments　results　show　that　the　proposed　method　matches　well　to　human　visual　perception　and　it　can　be　more　widely　applied　in　various　types　of　binary　images　such　as　text　images,　drawings,　handwritten　signature,　and　even　in　halftone　images.　©　2012　AICIT.