Image analysis – Image compression or coding
Reexamination Certificate
2000-03-20
2003-07-15
Tran, Phuoc (Department: 2621)
Image analysis
Image compression or coding
C382S244000, C382S239000, C382S245000, C382S233000, C382S248000
Reexamination Certificate
active
06594394
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to the field of seismic exploration, and more particularly, but without limitation, to an apparatus and method for efficiently processing seismic data sets to reduce data storage and transmission requirements.
BACKGROUND
Seismic exploration for oil, gas and other subterranean mineral deposits often begins with a seismic survey to identify suitable locations where such deposits might be found. These surveys are often carried out using acoustic wave techniques, wherein a wave source is employed to direct wave energy toward a surface of the earth (such as an ocean bottom). The wave energy travels downwardly into the earth and is reflected from various subterranean features back to a receiver array. The reflected wave energy provides information regarding the structure of various subterranean layers of rock, enabling decisions with regard to the desirability of further exploration at a given location.
By using multiple receivers arranged in a selected pattern, vast amounts of computerized, multi-dimensional image data can be obtained from the array. Depending upon the extent of the survey, the magnitudes of resulting seismic data sets can exceed several tens of terabytes (10
12
bytes) of data.
As will be appreciated, the storage, transmission and analysis of a data set of such magnitude can be extremely burdensome. Even with the continued advancements in satellite and internet data transmission techniques and computer processing and storage hardware and software solutions, the transmission of a multi-terabyte data set can take days, if not weeks, to complete. Moreover, the storage of such a data set can require a large number of magnetic storage devices (such as disc drives arranged in a RAID array). Data access and security backup are also substantial issues that require significant resources to address.
Accordingly, various data compression techniques have been employed in the art to reduce the storage and transmission requirements of large data sets, such as obtained during seismic surveys. For example, seismic data compression techniques have been recently discussed in Luo and Schuster, “Wave Packet Transform and Data Compression,” 62
nd
Annual Intern. Mtg. Soc. Expl. Geophys., 1992; Reiter and Heller, “Wavelet Transformation-Based Compression of NMO-Corrected CDP Gathers,” 64
th
Annual Intern. Mtg. Soc. Expl. Geophys., 1994; Donoho, Ergas and Villasenor, “High-Performance Seismic Trace Compression,” 65
th
Annual Intern. Mtg. Soc. Expl. Geophys., 1995; Vassiliou and Wickerhauser, “Comparison of Wavelet Image Coding Schemes for Seismic Data Compression,” 67
th
Annual Intern. Mtg. Soc. Expl. Geophys., 1997. See also U.S. Pat. No. 5,745,392 issued to Ergas et al., the general teachings of which are incorporated herein by reference.
The Ergas U.S. Pat. No. 5,745,392 reference is typical of prior art wavelet compression techniques, and generally teaches to 1) apply a wavelet transform to the data to generate subbands of transformed data in first, second and third dimensions; 2) apply uniform quantization to the subbands using integer equivalents representing ranges of values; 3) compress through lossless compression the redundant integer replacements; and 4) store the compressed data.
While operable, there are nevertheless limitations associated with these and other prior art approaches. For example, the Ergas U.S. Pat. No. 5,745,392 reference requires matrix transposition during the wavelet transformation, which is a computationally inefficient process to perform. Further, only approximate compression rates are obtained, so that each subset (gather) of the data can have a slightly different size (number of bytes), making difficult the exact identification of the beginning point for a selected gather. Moreover, at least some of the steps in the disclosed method are lossy, meaning that there is some degradation between the original data and the compressed data. While attempts are made to confine the majority of the losses to frequency ranges of little interest, substantial degradation of the decompressed data for high compression ratios in ranges of interest can nevertheless occur.
Accordingly, there is a continuing need for improvements in the manner in which large data sets, such as seismic surveys, are compressed and processed. It is to such improvements that the present invention is directed.
SUMMARY OF THE INVENTION
The present invention is directed to an apparatus and method for fast seismic data compression and a hardware/software integration platform for very fast data transmission through a data network.
The fast data compression method can achieve data compression ratios exceeding 100:1 for three dimensional (3-D) seismic data, with low noise degradation for subsequent seismic data processing of pre-stack data in the compressed domain or interpretation of post-stack seismic data. For lower compression ratios, such as from about 5:1 to about 20:1, the method produces essentially lossless compression through the use of a multi-layered data compression structure.
In accordance with a first aspect of the present invention, the data compression method generally comprises a fast multi-dimensional transform which uses an adaptive basis selection. Preferably, the transform is characterized as utilizing wavelet packets, local cosines, brushlets or other libraries of orthogonal/biorthogonal bases. Particularly, this step includes biorthogonal filter factorization, which occurs prior to filtering and reduces the computational time for a 1-D convolution by a factor of two, reduces the computational time for a 2-D convolution by a factor of four and a 3-D convolution by a factor of eight. Moreover, unlike the prior art the transform step requires no matrix transposition, significantly reducing computational complexity.
A second step in the method involves a novel coding approach which includes non-uniform scalar quantization of the wavelet packet transformed data and adaptive lossless run length coding of the thresholded quantized transform coefficients.
A third step involves repeating the first two steps over successive compression iterations to create a layered/hybrid/cascaded data compression algorithm. Thus, the different curves, surfaces, shapes and textures of the data are coded employing a different mathematical basis in each iteration. The quantization residuals are calculated at the end of each iteration and the first two steps are repeated as desired to achieve a preselected noise level or signal to noise (SNR) ratio.
In another aspect of the present invention, the data are decompressed by first determining how many compression iterations were employed in the compression process, after which the data are decompressed by applying a decoding step, which involves application of a run-length lossless decoding scheme, and an adaptive inverse transform step. These decoding and inverse transform steps are successively applied the same number of times as the original number of iterations during the compression operation. The decompressed data are thereafter combined with the residuals to output the decompressed data.
In another aspect of the present invention, a hardware/software scheme is employed to efficiently transmit and process data in a computerized network, such as a local area network (LAN) or a wide area network (WAN). A computer is dedicated as a decompression engine and is attached to the network using a high transfer rate interface (such as fibre-channel). A high capacity RAID is also attached to the network and stores the data which has been compressed as discussed above. A request for selected portions of the data causes the compression engine to select and decompress the data at a rate that meets or exceeds the available transfer rate of the interface, so that the decompression operation is transparent to the user.
REFERENCES:
patent: 4509150 (1985-04-01), Davis
patent: 4760563 (1988-07-01), Beylkin
patent: 4862167 (1989-08-01), Copeland, III
patent: 5084844 (1992-01-01), Wang
patent: 5384725 (1995
Averbuch Amir
Meyer Francois G.
Strömberg JanOlov
Vaddiliou Anthony A.
Alaui Amir
Brown, Jr. J. Robert
GeoEnergy, Inc.
Hunton & Williams LLP
Tran Phuoc
LandOfFree
Fast compression and transmission of seismic data does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Fast compression and transmission of seismic data, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Fast compression and transmission of seismic data will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3032426