wpc1.comp.dwpt.1d.lib

su 的源代码库

WPC1 - routines for compress a single seismic trace using wavelet packets 

wpc1Compress	compress a single trace
wpc1Uncompress	uncompress and generate a single trace
wpc1Numsmp	obtain the # of samples
wpc1Size	length of the compressed bit stream
wpc1Free	free up the space for the compressed data
wpc1Write	write the compressed data to a file
wpc1Read	read compressed data from a file 

Function Prototypes:
void *wpc1Compress(float *f, int n1, float error, int *nbytes);
int wpc1Uncompress(void *w1, float *f);
int wpc1Numsmp(void *wpc1);
int wpc1Size(void *wpc1);
void wpc1Free(void *wpc1);
void wpc1Write (void *wpc1, FILE *fp);
void *wpc1Read (FILE *fp);

wpc1Compress:
Input:
f	pointer to the trace
n1	# of samples in the trace
error	relative RMS error tolerable during compression

Output:
nbytes	# of bytes after compression 

Return:	pointer to the compressed data

wpc1Uncompress:
Input:
w1	pointer to the compressed trace

Output:
f	pointer to the uncompressed trace

Return:	consistency flag, 1 if wpc1Compressed data, 0 otherwise

wpc1Numsmp:
Input:
wpc1	pointer to the compressed data

Return:	# of samples per trace

wpc1Size:
Input:
wpc1	pointer to the compressed data

Return:	# of bytes after compression

wpc1Free:
Input:
wpc	pointer to the compressed data

wpc1Write:
Input:
wpc1	pointer to the compressed data
fp	file pointer to which to write

wpc1Read:
Input:
fp	file pointer to which to write

Return:	pointer to the compressed data

wpc1Compress:
Notes:
The compression here is done block by block, so the data is tiled 
first. For a single trace, to reduce the overhead, the shape gains
of each block are also quantized, but using 16bit, to maintain the
accuracy. Also, to reduce overhead, the last block (non-regular
block) is NOT transformed at all if it is shorter than half
of the tile size. If it is longer than half of the tile size, zero
padding is performed. An upgrade to the mixed-radix
version is expected soon. The size of the tiling is set in the 
routine wpcInitConfig. Those numbers are obtained empirically. 
Since we cannot determine the size of the output before compression,
memory is allocated within this routine, and adjusted after the
compression.  

Author:		Tong Chen, 08/04/94