A direct numerical simulation in turbulent curved channel flow is performed. The drifting Taylor-Gortler vortices are identified by applying a conditional averaging. A new algorithm is proposed based on the wavelet transform of the wall information. A continuous wavelet transform with Marr wavelets is employed to decompose the flow signals at a chosen length scale. An active cancellation is applied to attenuate the Taylor-Gortler vortices and to reduce the wall skin friction.