An　atomic　force　microscopy　(AFM)　image　is　acquired　by　probe　tip　scanning　on　the　surface　of　a　sample.　It　is　a　distorted　representation　of　the　sample　because　of　the　finite　size　of　the　tip.　To　modify　the　distorted　image　and　improve　the　measurement　accuracy　of　the　AFM　image,　it　is　important　to　estimate　the　tip　shape.　Tip　estimation　results　mainly　rely　on　the　sample-dimensional　uncertainty　and　AFM　image　noise.　More　reliable　data　of　tip　morphology　can　be　collected　if　there　is　a　suitable　tip　characteriser　to　reduce　the　sample-dimensional　uncertainty　and　improve　AFM　image　accuracy.　A　new　tip　characteriser　for　blind　reconstruction　of　AFM　tip　morphology　has　been　developed　through　the　fabrication　of　gold　nanorod　array　structures.　Based　on　template　synthesis,　the　gold　film　(surface　roughness,　162.1　nm)　was　deposited　on　porous　anodic　alumina　membrane　by　magnetron　sputtering.　The　well-ordered　nanorod　array　structures　(100-400　nm　height,　60　nm　diameter,　approximate　to　20　nm　at　apex　and　80　nm　pitch)　were　obtained.　In　combination　with　the　blind　reconstruction　algorithm,　the　prepared　nanostructures　were　used　as　the　tip　characteriser　to　estimate　the　morphology　of　the　traditional　AFM　Si3N4　probe,　which　can　effectively　reduce　the　influence　of　the　sample-dimensional　uncertainty　and　image　noise　on　the　result　of　the　tip　blind　reconstruction.