Most　of　existing　image　processing　algorithms　have　poor　real-time　performance　in　embedded　device　applications　and　occupy　too　much　software　resources.　In　this　paper,　a　fast　edge　detection　method　is　proposed　based　on　Zynq　platform.　The　xfOpenCv　acceleration　library　provided　by　Xilinx　is　used　to　implement　the　edge　detection　algorithm.　Specifically,　the　function　operation　process　is　accelerated　in　the　FPGA　through　the　acceleration　constraint　of　a　specific　function　implemented　in　the　xfOpenCv　library.　In　the　implementation　process,　the　SDx　development　environment　is　also　used　to　simplify　the　function　hardware　acceleration　development　process.　The　experimental　results　show　that　our　edge　detection　method　takes　less　time　than　other　methods,　and　has　better　real-time　performance　with　the　basically　same　detection　results.　©　2018　IEEE.