Optimizing the speckle noise for maximum efficacy of data acquisition in coherent imaging

Abstract

The impact of multiplicative speckle noise on data acquisition in coherent imaging is studied. This demonstrates the possibility to optimally adjust the level of the speckle noise in order to deliberately exploit, with maximum efficacy, the saturation naturally limiting linear image sensors such as CCD cameras, for instance. This constructive action of speckle noise cooperating with saturation can be interpreted as a novel instance of stochastic resonance or a useful-noise effect.

Tuning the Noise in Magnetic Resonance Imaging to Maximize Nonlinear Information Transmission

“…In most stochastic resonance studies, the information-noise mixture results from an additive coupling performed before the processing system. Other couplings have yet been also considered under the scope of stochastic resonance, like multiplicative noise [14,19,20] or phase noise [35][36][37]. An important motivation for these explorations is the physical existence of such mixtures in real world or by construction in existing technological devices.…”

“…The signal-noise coupling in MRI is neither additive nor multiplicative. To assess the specificity of this signalnoise coupling, we have chosen here for methodological reasons the same saturating static nonlinearity as used in [14,31] respectively for additive and multiplicative signal-noise mixture. With multiplicative signal-noise coupling, as shown in [14], in absence of noise, the output is generally carrying no information on the input image.…”

“…To assess the specificity of this signalnoise coupling, we have chosen here for methodological reasons the same saturating static nonlinearity as used in [14,31] respectively for additive and multiplicative signal-noise mixture. With multiplicative signal-noise coupling, as shown in [14], in absence of noise, the output is generally carrying no information on the input image. In the case of MRI signal-noise coupling, transmission of information is possible at a zero noise level and the stochastic resonance effect is therefore more similar to what is found [31] with additive signal-noise coupling where the signal has to be ill-positioned to transmit information at zero noise level to benefit from an injection of noise.…”

“…In addition, the static nonlinearity g [•] of Eq. (8) has already been studied in the context of stochastic resonance with additive coupling [31] occurring in thermal noise or with multiplicative coupling occurring with speckle noise in optics [14] for instance. As given in Eq.…”

“…Stochastic resonance, since its introduction [6] in the domain of nonlinear physics, has gradually emerged as a general paradigm of noise-assisted information processing, reported to operate in many different areas such as electronic circuits, lasers, neurons, nanodevices or chemical reactions [5,7]. Stochastic resonance has mainly been reported with mono-dimensional signals for various digital processes (including for instance quantization [8] detection [9] or estimation [10]) and its application to images is more recent [11][12][13][14][15][16][17]. Also, stochastic resonance has essentially been explored with additive signal-noise coupling [7], and less frequently with multiplicative signal-noise coupling [14,[18][19][20].…”