Linear prediction coding for compressing of seismic data

Claim 1

A method of seismic exploration, comprising the steps of transmitting acoustic energy through the subterranean structure of the earth to be reflected from an interface within the earth and return upwardly to the surface of the earth, detecting return of said energy, outputting electrical data signals in response thereto, and compressing said data signals for storage or transmission, said compression step comprising the following steps: sampling said signals at a fixed rate to produce data samples; generating a prediction operator; analyzing pluralities of said samples successively in accordance with said prediction operator to yield a predicted value for the next data sample; comparing said next data sample with the corresponding predicted value to yield a prediction error value; quantizing said prediction error value to yield a quantum number; and storing or transmitting said quantum number; wherein said step of generating the prediction operator includes performing autocovariance on a plurality of samples, whereby low-frequency drift of said prediction errors is avoided.

Claim 2

A method of seismic exploration, comprising the steps of transmitting acoustic energy through the subterranean structure of the earth to be reflected from an interface within the earth and return upwardly to the surface of the earth, detecting return of said energy, outputting electrical data signals in response thereto, and compressing said data signals for storage or transmission, said compression step comprising the following steps: sampling said signals at a fixed rate to produce data samples; generating a prediction operator; analyzing pluralities of said samples successively in accordance with said prediction operator to yield a predicted value for the next data sample; comparing said next data sample with the corresponding predicted value to yield a prediction error value; quantizing said prediction error value to yield a quantum number; and storing or transmitting said quantum number; wherein said step of generating the prediction operator includes performing autocovariance on a plurality of samples, whereby low-frequency drift of said prediction errors is avoided.

Claim 8

A method of compression of samples of seismic signals, comprising the following steps: generating a prediction operator by analysis performed with respect to a predetermined number of said samples; applying said operator to a series of said samples of predetermined length to yield a predicted value for a next sample; measuring the difference between said predicted value and the next sample; comparing said difference to a series of numbered ranges of variable size; recording the number of the range within which said difference fits; varying the size of said ranges with respect to said number; and removing the mean values of said predetermined number of samples prior to generation of said prediction operator, so that low-frequency drift of said measured differences is avoided.

Claim 9

A method of compression of samples of seismic signals, comprising the following steps: generating a prediction operator by analysis performed with respect to a predetermined number of said samples; applying said operator to a series of said samples of predetermined length to yield a predicted value for a next sample; measuring the difference between said predicted value and the next sample; comparing said difference to a series of numbered ranges of variable size; recording the number of the range within which said difference fits; varying the size of said ranges with respect to said number; and removing the mean values of said predetermined number of samples prior to generation of said prediction operator, so that low-frequency drift of said measured differences is avoided.

Claim 10

A method of seismic exploration of the earth comprising the steps of: transmitting acoustical energy into the earth to be detected from interfaces within the subterranean structure of the earth; detecting return of said energy to the surface of the earth and outputting electrical data signals in response thereto; sampling said electrical signals at regular intervals; and processing said sampled signals to generate a pictorial representation of the subterranean structure of the earth, wherein said samples are compressed prior to storage or transmission and decompressed prior to generation of said pictorial representation of the subterranean structure of the earth, said compression including the following steps: In a first SET UP mode: selection of a series of said samples comprising a predetermined number of said samples; removal of the mean values of said samples from said series of samples; and generating a prediction operator using said series of samples having had the mean values removed therefrom; and In a subsequent RUN mode: utilizing said prediction operator to generate predicted values for successive samples; measuring the difference between actual received successive samples and the predicted values therefor; characterizing said difference by comparison thereof with possible ranges of difference values; identifying the one of said ranges within which said difference value falls; and storing or transmitting the identification of the range within which each difference value falls, as the compressed version of the corresponding data sample, whereby substantial economies in said storage or transmission are achieved.

Claim 11

A method of seismic exploration of the earth comprising the steps of: transmitting acoustical energy into the earth to be detected from interfaces within the subterranean structure of the earth; detecting return of said energy to the surface of the earth and outputting electrical data signals in response thereto; sampling said electrical signals at regular intervals; and processing said sampled signals to generate a pictorial representation of the subterranean structure of the earth, wherein said samples are compressed prior to storage or transmission and decompressed prior to generation of said pictorial representation of the subterranean structure of the earth, said compression including the following steps: In a first SET UP mode: selection of a series of said samples comprising a predetermined number of said samples; removal of the mean values of said samples from said series of samples; and generating a prediction operator using said series of samples having had the mean values removed therefrom; and In a subsequent RUN mode: utilizing said prediction operator to generate predicted values for successive samples; measuring the difference between actual received successive samples and the predicted values therefor; characterizing said difference by comparison thereof with possible ranges of difference values; identifying the one of said ranges within which said difference value falls; and storing or transmitting the identification of the range within which each difference value falls, as the compressed version of the corresponding data sample, whereby substantial economies in said storage or transmission are achieved.