Shape　from　focus　(SFF)　is　a　technique　for　reconstructing　the　3D　shape　of　a　scene　from　a　sequence　of　multi-focus　images.　The　quality　of　the　shape　reconstruction　depends　on　the　accuracy　of　the　initial　focus　volume　(FV)　and　the　optimization　of　the　initial　depth　map.　The　initial　FV　is　calculated　by　the　focus　measure　(FM)　operator.　Traditional　FM　operators　perform　poorly　in　preserving　structural　edges　and　weak　texture　areas.　In　depth　map　optimization,　strong　smoothness　constraints　are　mainly　used　for　improvement,　but　this　relies　on　the　accuracy　of　the　initial　depth　map.　Therefore,　in　this　paper,　we　propose　a　complete　SFF　method　framework.　We　suggest　performing　detail　transformation　on　the　images　to　enhance　texture　information,　especially　in　weak　texture　areas.　For　micro-images,　a　curvature　method　is　proposed　to　evaluate　the　sharpness　level　of　the　images.　Noise　in　the　initial　depth　map　is　divided　into　two　types　and　optimized　using　spatial　clustering　and　static　guided　filtering　methods.　Experiments　have　been　conducted　to　validate　the　effectiveness　of　the　proposed　method　using　both　synthetic　and　real　objects.　Compared　with　existing　methods,　the　proposed　method　achieves　excellent　results　in　accurately　reconstructing　3D　shapes.　©　2023