An　algorithm　for　autonomous　selection　of　optimal　landing　site　based　on　the　image　is　proposed.　Firstly,　referenced　rocks＇　shadow　and　light　areas　are　extracted　by　the　use　of　double　threshold　segmentation　based　on　the　two-dimensional　maximum　entropy.　Then,　the　threshold　segmentation　image　is　processed　further,　including　the　combination　of　the　shadow　and　light　areas　in　order　to　form　obstacle　areas　and　extracting　connected　areas　etc.　However,　the　selection　of　the　safe　landing　areas　should　not　only　take　the　safety　into　consideration,　but　also　pay　more　attention　to　the　value　of　scientific　research.　Therefore,　the　traditional　method　should　be　improved.　Finally,　the　optimal　landing　site　is　selected　based　on　fuzzy　rules　merged　expertise.　Experiments　demonstrate　that　the　algorithm　can　be　used　to　select　the　optimal　landing　site　accurately　and　it　proves　that　the　method　proposed　is　feasible　and　robust.