A　method　with　adaptive　dual-threshold　energy　and　cyclostationary　feature　detection　is　presented,　which　aims　at　reducing　high　false　alarm　rate　and　computation　complexity　occurred　in　the　single　threshold　energy　detection　algorithm　and　obtaining　the　less　radio　channel　average　capture　time　and　lower　signal　detection　computational　complexity.　The　two　thresholds　are　adaptively　adjusted　to　search　the　spectrum　based　on　the　channel　status　of　occupancy,　the　wireless　signal　received　in　radio　microphone　is　send　to　the　first　integrator,　and　the　signal　outside　dual-threshold　region　is　detected　using　the　energy　detection　method.　The　output　of　the　energy　accumulated　in　the　first　comparator　less　than　the　low-threshold　shows　no　wireless　signal　be　detected,　the　second　integrator　works　if　the　output　of　energy　accumulated　exceed　the　low-threshold,　when　the　energy　accumulated　in　second　comparator　exceed　upper-threshold,　the　third　comparator　starts　to　judge,　the　false　alarm　state　is　detected　if　the　energy　exceed　the　synchronous　threshold,　the　wireless　microphone　signal　is　detected　using　the　cyclostationary　feature　detection　outside　the　dual-threshold,　and　the　guardband　is　properly　set　up,　it　permits　the　unauthorized　users　to　make　full　use　of　spectrum　occured　outside　the　guard　band.　The　simulation　results　show　that,　it　is　effectively　using　the　detection　algorithm　proposed　to　improve　signal　detection　efficiency　and　spectrum　sensing　accuracy　in　wireless　cognitive　networks.