Application of William H. Cox, 342 F.2d 1017 (C.C.P.A. 1965)Annotate this Case
George A. Smith, Philadelphia, Pa., for appellant.
Clarence W. Moore, Washington, D. C. (Jere W. Sears, Washington, D. C., of counsel), for Comr. of Patents.
Before WORLEY, Chief Judge, and RICH, MARTIN, SMITH and ALMOND, Judges.
Appealed claims 9-13 of appellant's application serial No. 823,772, filed June 29, 1959, relate to "Analysis of Seismic Records" and are drawn to a new use for an old apparatus. While the precise statutory ground of rejection is not easily discernible from the examiner's answer, the board appeared to consider the rejection as one for obviousness under 35 U.S.C. § 103. Both the solicitor and appellant seem to agree that section 103 is involved and we will proceed on that basis.
In the art of seismic exploration a persistent problem is that the detected signal resulting from a seismic "shot" contains large amounts of noise such as wind through trees, traffic sounds, electrical noise and other irrelevant information. The result is that the pertinent reflected signal which provides the sought-for data to the seismic explorers is obscured and difficult to recognize. A common approach to the problem has been the utilization of filters designed to suppress all frequencies in the detected composite pattern except those corresponding to the pertinent reflected signal. Appellant asserts in his specification, however, that the frequency-filtering approach is not wholly satisfactory due to certain limitations in filter design with respect to frequency response, phase shift and transient properties.
Appellant's specification and claims are directed to a new approach to the problem of signal discrimination and recovery which involves recovering the pertinent signal on the basis of discrimination between wave shapes rather than frequencies alone. In rather simplified form, appellant's method is as follows:
Using known techniques, appellant computes theoretically the waveform which could be expected to be reflected from a particular subsurface conformation. Once he has determined this theoretical "wavelet," he uses it in apparatus capable of continuously multiplying the ordinates of the "wavelet" by the ordinates of an actual, noise-containing seismic signal, and concurrently integrating the products over the time interval of the "wavelet." Such operation results in relative enhancement of the pertinent reflected signal, so that the trained observer is much more readily able to recognize the pertinent signal when it occurs in the complex wave pattern. Repeating the operation a number of times for comparison purposes serves to identify spurious signals resulting from chance correspondence of noise and "wavelet" peaks.
As one of the physical embodiments appellant discloses an optical arrangement said to be capable of performing an actual continuous integration. The preferred embodiment, however, is an electrical system which utilizes a delay line and attenuators. The delay line serves to isolate the ordinates of the actual detected signal waveform, while the attenuators are set in accordance with the ordinates of the theoretical wavelet. Such an arrangement cannot perform a true integration, but only an approximation thereof in terms of a summation of discrete values. The delay line may be electrical, or mechanical or acoustic; the delay may also be accomplished by means of staggered pick-ups on a multi-channel magnetic tape.
Claim 13 is illustrative of those on appeal and reads:
"13. The new use, for analysis of a seismic waveform varying with time, of an apparatus comprising a tapped delay line providing simultaneously a plurality of outputs each of which represents the value of a waveform at a predetermined different instant of time, means comprising a plurality of attenuators individual to said outputs receiving said outputs and multiplying each by a predetermined individual constant to provide product outputs, means receiving said product outputs and summing them to provide a summed output and means for recording said summed output as a function of the time scale of said waveform, said new use involving introducing a seismic waveform to the apparatus as the waveform involved and adjusting said attenuators to provide as each of said individual constants the amplitude of a corresponding instantaneous portion of an expected reflection of a seismic wave."
The Patent Office relied upon the following patents as prior art:
Yost 2,794,965 June 4, 1957 Calvert et al. 2,801,351 July 30, 1957
Calvert et al. was cited to meet the specific "tapped delay line" limitation of claims 12 and 13 and need not be considered by us, since it is our view that the claims stand or fall together.
Yost discloses apparatus and method for accomplishing the same ultimate result as appellant, i. e., easier identification of a pertinent reflection from a composite seismic trace. In essence, and stated very simply, Yost's method amounts to establishing a representative noise waveform and subtracting it from the composite waveform to leave the pertinent reflected signal standing alone. Yost's apparatus is in many ways similar to that disclosed by appellant, and we will assume arguendo that Yost's equipment would be suitable for use in appellant's claimed method with but minor modifications.
Thus the fundamental difference between the method of Yost and appellant's claimed method is clear: Yost establishes an expected noise waveform and subtracts it from the actual composite waveform in order to isolate the pertinent reflected signal. Appellant establishes the waveform of an expected reflected signal and combines it with the actual composite waveform in such a way as to augment the pertinent reflected signal.
Although the board apparently recognized this difference, it regarded appellant's use of a theoretical wavelet rather than a noise waveform as merely "arbitrary." We think such a characterization of what is the very essence of appellant's claimed new use is erroneous. Indeed, it is the least arbitrary feature of the claimed method, since it depends entirely upon the theoretical shape which could be expected for a reflected signal in a particular geological area. And certainly it cannot be seriously argued that the concept of using a theoretical wavelet rather than a noise waveform was "arbitrary." It was, as are most unobvious innovations, the product of careful observation of prior art drawbacks and thoughtful reflection on means of overcoming them, with an ultimate reasoned (and not arbitrary) selection of a particular means.
We agree with appellant that establishing a representative noise waveform, as required by Yost, would in general be much more difficult than establishing the shape of an expected reflection. The solicitor seems also to agree, for he states: "* * * Yost's approach requires the selection of a truly representative noise wave * * *. This may be difficult in view of the complex nature of the noise wave * * *." We think this factor emphasizes the unobviousness of appellant's method. For if it would have been obvious to add a theoretical reflection rather than subtract noise, we think the desirability of such an expedient makes it probable that Yost would have done so. But we have searched Yost in vain for an indication of any use of a theoretical wavelet instead of a noise pattern.1
On balance, we believe the board erred in concluding that there is no patentable distinction between the method disclosed by Yost and that claimed by appellant. While it is true that both methods accomplish the same ultimate result, and may possibly utilize the same apparatus, the paths by which the goal is reached are conceptually very different, and we think appellant's path is the easier to travel. This factor, coupled with the absence of any recognition in Yost that alternative paths were possible, compels the conclusion that appellant's method would not have been obvious at the time he invented it. The appealed decision is accordingly reversed.
The only statement in Yost of even remote pertinence is the following:
"* * * In a similar manner the anomalous or reflection components characteristic of a given area are characterized in a second auto-correlation function * * * which is obtained by similarly operating on a representative reflection record obtained in the same geological area. * * *" What this sentence seems to say, in effect, is that the waveform of a pertinent reflected signal could be determined in the same way as the noise waveform. But there is no suggestion that the former could or should be used in the same way as the latter. We do not agree with the solicitor's conclusion that subtraction of the expected reflected signal instead of noise so as to indicate the presence of actual reflections as flat valleys among noise peaks, would be "an obvious extension of Yost's signal cancelling idea to the balance of his seismographic trace." Moreover, if such an "extension" were obvious, even then it would not necessarily follow that appellant's product integration scheme would be patentably indistinguishable from a subtraction scheme.