Baseline　correction　in　time　series　of　ground　motion　is　a　challenging　problem　in　research　pertaining　to　strong　ground　motion.　In　previous　studies　in　this　area,　numerous　methods　have　been　proposed　to　identify　baseline　drift.　However,　there　is　no　systematic　approach　to　verify　the　proposed　methods.　In　this　paper,　we　propose　a　vision-based　approach　to　evaluate　baseline　correction　methods　quantitatively.　In　addition,　for　our　proposed　baseline　correction　method　based　on　L1-norm　optimization,　we　analyze　the　method　in　detail　and　provide　a　thorough　analysis　of　the　method　based　on　experimental　data.　From　the　comparison,　our　method　shows　good　performance　in　recovering　the　baseline.