The elegance and efficacy of optical methods rest upon their utilization of beams of photons as the finest available means of probing the objects under study. It is convenient to view optical observations and measurements as interactions among object, instruments, and subject, resulting in the transfer of a finite amount of information. Thus, the optical evaluation of instruments and methods becomes a search for the best way of effecting this transfer.A survey of several basic theorems of geometric optics includes Fresnel's construction, Babinet's theorem of complementary diaphragms, de Fermat's principle, and the klalus theorem. The present study points to the need to supplement the deficiencies of the geometric approach. One of the relatively simple w a y s of achieving this objective utilizes a method of spatial frequency analysis. Thus, every optical system transfer function can be characterized by a finite frequency-response characteristic. It is shown that simple relations exist between the Shuster slip factor, the Rayleigh resolution, effective slit width, fnumber and numerical aperture, and the cut-off frequency at the resolution limit of linear noisy optical communication channels considered as low band-pass frequency filters. These considerations lead to the formulation of performance criteria of a statistical nature that are quite readily accessible to experimentation (relative image structural content, fidelity factor, Strehl definition, correlation factor, and O'Neill sharpness factocy). The practical u s e of these new concepts is facilitated by the introduction of the sampling theorem, which permits the prediction of the best filtering to be used in each case (such a s periodic objects, transients, and random objects) The analogies between optical and electronic channels lead to an interpretation of optical noise as the sum of all the aberrations plus transient disturbances, Contemporary methods of image-frequency filtration and evaluation (such a s apodization, coating functions, twewave-length reimaging, and amplitude contrast process) are briefly reviewed from the parallel standpoints of a general interaction theory, the Shannon theorem of communication, and Ingelst&'s theorem of optical indeterminacy. The latter theorem probably opens the best way for a more complete interpretation of the optical observation of biological objects, *This paper, illustrated with slldes, was presented at a meeting Of the Divldon on January 20, 1959. It will be published In the Annala of The New Ycnk Academy of Sciences.
324
