This　paper　presents　a　three-dimensional　(3D)　sound-field　reproduction　(SFR)　method　based　on　the　Pressure-Matching　(PM)　approach.　The　desired　sound　field　is　modeled　to　be　reverberation　field　to　approximate　a　real　conference　room　through　the　image-source　model　and　the　reflection　model　of　the　incoherent　rough　room　surfaces.　The　whole　reproduction　procedure　can　be　divided　into　two　parts.　At　the　first　step,　a　large　number　of　candidate　loudspeaker　positions　for　selecting,　and　the　most　efficient　active　loudspeaker　is　chosen　based　on　the　Alternating　Direction　Method　of　Multipliers　(ADMM)　based　Lasso.　During　the　second　step,　all　the　selected　active　loudspeaker　weights　are　calculated　via　regularized　Least-Square　(LS)　method.　Simulation　results　shows　that　the　proposed　SFR　approach　outperforms　the　existing　PM-based　SFR　method　in　most　cases　for　reverberation　environment,　and　the　runtime　is　far　less　than　the　reference　methods,　which　means　that　the　proposed　approach　is　especially　suited　for　low-latency　sound　field　synthesis　in　reverberant　rooms.　©　2019,　Springer　Nature　Switzerland　AG.