Optical　navigation　technology　is　widely　used　in　precision　guidance　for　autonomous　landing　and　obstacle　avoidance.　According　to　the　requirements　of　the　computer　vision　landing　location　system　on　a　planetary　lander,　this　paper　mainly　studies　the　image　feature　matching　algorithm　in　the　process　of　autonomous　navigation　during　landing　on　asteroids.　A　matching　algorithm　based　on　local　feature　regions　was　proposed,　which　takes　the　invariant　moment　and　mutual　information　as　similarty　measures　in　the　two　phases　of　feature　matching,　and　introduces　bidirectional　matching　strategy,　to　improve　matching　accuracy.　In　order　to　solve　the　problems　of　the　feature　matching　for　zoom　image,　this　paper　put　forward　an　improved　SIFT　algorithm,　the　experimental　results　showed　that　this　method　can　not　only　match　feature　points　effectively　for　zoom　image,　but　also　achieve　high　matching　accuracy　and　is　qualified　for　engineering　application.　©　2014　IEEE.