Shannon information storage in noisy phase-modulated fringes and fringe-data compression by phase-shifting algorithms.

Optical phase-modulated fringe-patterns are usually digitized with XxY pixels and 8 bits/pixel (or higher) gray-levels. The digitized 8 bits/pixel are raw-data bits, not Shannon information bits. Here we show that noisy fringe-patterns store much less Shannon information than the capacity of the digitizing camera. This means that high signal-to-noise ratio (S/N) cameras may waste to noise most bits/pixel. For example one would not use smartphone cameras for high quality phase-metrology, because of their lower (S/N) images. However smartphones digitize high-resolution (12 megapixel) images, and as we show here, the information storage of an image depends on its bandwidth and its (S/N). The standard formalism for measuring information are the Shannon-entropy H, and the Shannon capacity theorem (SCT). According to SCT, low (S/N) images may be compensated with a larger fringe-bandwidth to obtain high-information phase measurements. So broad bandwidth fringes may give high quality phase, in spite of digitizing low (S/N) fringe images. Most real-life images are redundant, they have smooth zones where the pixel-value do not change much, and data compression algorithms are paramount for image transmission/storage. Shannon's capacity theorem is used to gauge competing image compression algorithms. Here we show that phase-modulated phase-shifted fringes are highly correlated, and as a consequence, phase-shifting algorithms (PSAs) may be used as fringe-data compressors. Therefore a PSA may compress a large number of phase-shifted fringes into a single complex-valued image. This is important in spaceborne optical/RADAR phase-telemetry where downlink is severely limited by huge distance and low-power downlink. That is, instead of transmitting M phase-shifted fringes, one only transmit the phase-demodulated signal as compressed sensing data.