纺织品检测中的模式识别应用

将模式识别方法用于毛巾和纺织面料生产过程中的瑕点检测, 研究了模糊小波模式识别方法, 对毛巾生产过程的多种瑕点监测进行了算法分析和简要论述, 这种算法具有更强的实用性和鲁棒性. 又由于系统采用DSP实现, 使识别速度大大提高, 完全能满足实时性的要求.

A way of pattern recognition for identification in textile

An effective method of pattern recognition is introduced. It is used in towel and textile fabric making process. For the multi-feature extraction using the fuzzy wavelets intelligent arithmetic and making FWA analysis, it combines fuzzy tools and wavelet transform techniques for providing a robust feature extraction and failure detection and identification scheme. The input signal first undergoes preprocessing and then the features are extracted using the wavelet transform. The extracted features are fuzzified and an inference engine uses the knowledeg-base to declare fault conditions. The fuzzification process adapts dynamically to external disturbances so that the classification performance is continuously improved. The architecture can be used in practical field for feature extraction and defect identification in textile fabric. The detection speed is quickened and real-time operation is satisfied.