Image　threshold　segmentation　based　on　entropy　is　classical　method.　The　time　cost　of　applying　the　two-dimensional　maximum　entropy　and　enumeration　threshold　segmentation　method　is　unacceptable,　so　that　the　genetic　algorithms　is　adopted　to　improve　efficiency.　Because　of　the　premature　convergence　of　traditional　genetic　algorithm,　the　performance　of　image　threshold　segmentation　is　constrained.　We　propose　a　2-D　maximum　entropy　threshold　segmentation　method　based　on　the　auxiliary　individual　oriented　crossover　genetic　algorithm　(AIOXGA)　to　improve　the　speed　and　success　rate　of　image　threshold　segmentation.　The　introduction　of　the　AIOX　operator　reduces　the　blindness　of　the　genetic　algorithm　and　improves　the　optimization　efficiency.　This　method　was　compared　with　enumeration　method,　standard　genetic　algorithm　and　original　oriented　genetic　algorithm(OGA)　in　image　segmentation　experiments.　The　results　show　that　the　performance　of　this　method　is　better　than　that　of　traditional　methods.　©　2019　IEEE.