In　this　paper,　it　is　researched　that　a　single　focus　measure　operator　is　unable　to　apply　to　different　texture　types,　making　it　difficult　to　obtain　accurate　depth　maps　and　unable　to　achieve　three-dimensional　shape　reconstruction　of　microscopic　objects.　However,　another　disadvantage　is　that　real　object　images　often　have　different　contrasts,　which　can　also　lead　to　errors　in　depth　map　calculations　and　result　in　poor　performance.　In　order　to　address　this　issue,　this　paper　proposes　the　use　of　multi-confidence　measure　in　combination　with　PCA　for　the　focus　measure　operator,　resulting　in　a　more　accurate　depth　map.　Tests　were　conducted　on　synthetic　and　real　objects　to　evaluate　the　proposed　approach,　and　the　experimental　results　showed　that　our　algorithm　outperformed　traditional　focus　measurement　algorithms,　obtaining　more　accurate　depth　maps.　　©　2023　IEEE.