An　accurate　navigation　system　is　an　essential　and　important　part　for　many　applications　carried　out　in　the　indoor　environments.　In　the　absence　of　absolute　positioning　information　such　as　global　positioning　system,　the　navigation　solution　which　relies　on　previous　system　states　such　as　dead　reckoning　has　shown　disadvantage　over　long　time　running.　In　contrast,　we　introduce　an　novel　and　inexpensive　sensor　fusion　based　approach　to　solve　the　problem.　The　key　idea　is　to　use　visual　gyroscope　as　a　complementary　source　for　system　heading　estimation　since　it　only　depends　on　the　scene　observed　by　the　camera.　The　introduced　methods　only　requires　a　Kinect　and　a　low　cost　inertial　measurement　unit.　By　using　the　same　mechanism,　the　introduced　method　has　been　tested　for　the　applications　of　both　indoor　mobile　robot　and　pedestrian　navigation.　Field　experiments　have　been　carried　out　and　corresponding　results　are　presented.