sucwt.su.main

su 的源代码库

 SUCWT - generates Continous Wavelet Transform amplitude, regularity	
         analysis in the wavelet basis					

     sucwt < stdin [Optional parameters ] > stdout			

 Required Parameters:							
 none									

 Optional Parameters:							
 base=10	Base value for wavelet transform scales			
 first=-1	First exponent value for wavelet transform scales	
 expinc=0.01	Exponent increment for wavelet transform scales		
 last=1.5	Last exponent value for wavelet transform scales	

 Wavelet Parameters:							
 wtype=0		2nd derivative of Gaussian (Mexican hat)	
			=1 4th derivative of Gaussian (witch's hat)	
			=2 6th derivative of Gaussian (wizard's hat)	
 nwavelet=1024		number of samples in the wavelet		
 xmin=-20		minimum x value wavelet is computed		
 xcenter=0		center x value  wavelet is computed 		
 xmax=20		maximum x value wavelet is computed		
 sigma=1		sharpness parameter ( sigma > 1 sharper)	

 verbose=0		silent, =1 chatty				
 holder=0		=1 compute Holder regularity estimate		
 divisor=1.0		a floating point number >= 1.0 (see notes)	

 Notes: 								
 This is the CWT version of the time frequency analysis notion that is 
 applied in sugabor.							
 The parameter base is the base of the power that is applied to scale	
 the wavelet. Some mathematical literature assume base 2. Base 10 works
 well here.								

 Default option yields an output similar to that of sugabor. With the  
 parameter holder=1 an estimate of the instantaneous Holder regularity 
 (the Holder exponent) is output for each input data value. The result 
 is a Holder exponent trace for each corresponding input data trace.	

 The strict definition of the Holder exponent is the maximum slope of  
 the rise of the spectrum in the log(amplitude) versus log(scale) domain:

 divisor=1.0 means the exponent is computed simply by fitting a line   
 through all of the values in the transform. A value of divisor>1.0    
 indicates that the Holder exponent is determined as the max of slopes 
 found in (total scales)/divisor length segments.			

 Some experimentation with the parameters nwavelet, first, last, and   
 expinc may be necessary before a desirable output is obtained. The	
 most effective way to proceed is to perform a number of tests with    
 holder=0 to determine the range of first, last, and expinc that best  
 represents the data in the wavelet domain. Then experimentation with  
 holder=1 and values of divisor>=1.0 may proceed.			



 Credits: 
	CWP: John Stockwell, Nov 2004
 inspired in part by "bhpcwt" in the BHP_SU package, code written by
	BHP: Michael Glinsky,	c. 2002, based loosely on a Matlab CWT function

 References: 
         
 Li C.H., (2004), Information passage from acoustic impedence to
 seismogram: Perspectives from wavelet-based multiscale analysis, 
 Journal of Geophysical Research, vol. 109, B07301, p.1-10.
         
 Mallat, S. and  W. L. Hwang, (1992),  Singularity detection and
 processing with wavelets,  IEEE Transactions on information, v 38,
 March 1992, p.617 - 643.