Multi-scale Abnormal Change Detection And Crack Detection

The detection of abrupt changes is an effective method of detecting faint changes of signal by predicting change regularity of signal and finding the part of abrupt changes of signal.In recent, it is spread out in automatic control,time series analysis and industiral quality management,and it is paied attention to in the theoretic and industrial communities.This paper introduced multiscale abrupt change detection by combinating abrupt change detection and multiscale wavelet transform.It can detect the abrupt changes of signal in the different scales,and it develops the theory of abrupt change detection.We gain some better result by combinating this method and fracture detection.This paper gains some production and contribution,and it includes:1. Because multiscale abrupt change detection is similar with multiscale edge detection in the theory and method,the method of multiscale abrupt change detection uses the multiscale edge detection technique by combinating abrupt change detection and multiscale wavelet analysis;2. By putting partial wavelet coefficient in wavelet transform based on slip time window,we gain a fast wavelet transform algorithm that pays out a few storage space.The theory calculation result shows that the fast wavelet transform algorithm based on slip time window is fast and stable,and it can be used effectively and widely;3. We introduce multiscale abrupt change detection technique in fracture recognition.This method displays multiscale abrupt change detection technique of seismic fracture,and offers a new thought and method for estimating the geological lateral variances in the earth,for predicting fracture system research.Real application shows that this method is effective.