©　2016,　Editorial　Office　of　the　Journal　of　Combustion　Science　and　Technology.　All　right　reserved.Adaptive　threshold　binarization　was　utilized　to　track　the　turbulent　flame　front　and　its　algorithm　was　introduced　in　this　paper.The　image　intensity　histogram　combined,local　threshold　was　selected　based　on　local　pixel　intensity.And　noise　was　eliminated　by　seed-mediated　growth　to　obtain　the　legible　flame　front.The　processing　results　of　CH4/air　flame　image　by　the　proposed　method,traditional　binarization　and　the　front　tracking　of　Matlab　software　under　different　ambient　pressure　were　compared.Results　show　that　the　flame　fronts　are　seriously　vague　using　the　traditional　binarization　and　the　front　tracking　of　Matlab　when　the　signal-to-noise　ratio　is　low,while　continuous　flame　front　can　still　be　acquired　by　the　proposed　method.