In　order　to　solve　the　problem　of　large-scale　image　retrieval　for　complex　images,　this　paper　presents　a　novel　ROI-based　image　retrieval　algorithm.　In　this　algorithm,　a　new　color　image　segmentation　algorithm　based　on　information　mutation　is　used　to　divide　an　image　into　several　target　areas　firstly;　then　the　features　of　image　regions　are　extracted;　finally,　a　new　similar　measurement　criterion　is　defined.　In　the　process　of　image　retrieval,　the　region　of　interest　is　selected　by　user,　and　then　the　ROI-based　image　retrieval　is　accomplished　according　to　an　ergodic　searching　method.　Experimental　results　show　that　the　new　algorithm　presented　in　this　paper　has　the　advantage　of　high　efficiency　and　is　robust　to　image　resizing　and　translation.　It　could　make　use　of　the　color　and　spatial　information　of　an　image　effectively,　and　the　results　hold　favorable　consistency　in　terms　of　human　perception.　©　2007　IEEE.