The　implied　volatility　in　the　Black-Scholes　model　is　assumed　as　a　constant.　However,　empirical　analysis　proves　that　the　values　of　the　implied　volatility　based　on　the　same　underlying　asset　vary　with　the　underlying　asset　price,　maturity　date,　time　to　maturity,　and　the　strike　price.　Ronald　Lagnado　and　Stanley　Osher　and　Chiarella　proposed　the　techniques　for　calibrating　derivative　pricing　by　solving　an　inverse　problem.　In　this　paper,　an　improved　model　is　proposed.　In　which,　the　influence　of　the　options　with　different　strike　prices　is　distinguished　by　weights　and　the　choices　of　the　weight　functions　are　discussed.　The　approach　we　take　is　numerically　solving　an　inverse　problem　which　is　more　solid　in　theory　than　simple　regression　methods.　Numerical　results　show　that　our　model　has　better　adaptivity　and　accuracy　than　traditional　models.