In the process of the textile industry,the work flow of the circular weft knitting machine is finished by the manual work,the work intensity is big,the error rate is high,and the product quality and production efficiency are affected.According to the automatic identification and localization of the curtain cloth area,the conclusion and criterion of the FFT transform based on one dimensional in static state can be effectively identified and detected.Based on one dimensional FFT transform,this paper studies and analyzes the frequency conversion of high speed moving images,and the static recognition algorithm can not be effectively identified.A new recognition method is proposed,which is used to increase the filtering function,to extract the local features,to analyze the dynamic threshold,to increase the robustness of the algorithm,to reduce the computational complexity of the algorithm,and to improve the reliability of the recognition.The method is proved to be feasible and reliable,and it can be applied to the production line of the cutting machine through the VC++software programming.