Agnès Delahaies

Structural Similarity Measure to Assess Improvement by Noise in Nonlinear Image Transmission

We show that the structural similarity index is able to register stochastic resonance or improvement by noise in nonlinear image transmission, and sometimes when not registered by traditional measures of image similarity, and that in this task this index remains in good match with the visual appreciation of image quality.

Local-feature-based similarity measure for stochastic resonance in visual perception of spatially structured images.

For images, stochastic resonance or useful-noise effects have previously been assessed with low-level pixel-based information measures. Such measures are not sensitive to coherent spatial structures usually existing in images. As a result, we show that such measures are not sufficient to properly account for stochastic resonance occurring in visual perception. We introduce higher-level similarity measures, inspired from visual perception, and based on local feature descriptors of scale invariant feature transform (SIFT) type. We demonstrate that such SIFT-based measures allow for an assessment of stochastic resonance that matches the visual perception of images with spatial structures. Constructive action of noise is registered in this way with both additive noise and multiplicative speckle noise. Speckle noise, with its grainy appearance, is particularly prone to introducing spurious spatial structures in images, and the stochastic resonance visually perceived and quantitatively assessed with SIFT-based measures is specially examined in this context.

Joint acquisition-processing approach to optimize observation scales in noisy imaging

In imaging, the choice of an observation scale is conventionally settled by the operator in charge of the image acquisition, who is left alone with tuning the framing and zooming parameters of the imaging system. In a somewhat decoupled manner, the operator in charge of processing the data has access to the images after their acquisition, and seeks to extract information from the observed scene. This Letter proposes a manifestation of the interest of an alternative joint acquisition-processing approach. We demonstrate with quantitative informational measures how the choice of an observation scale can be directly related to the performance of the final information processing task. Illustrations are given with various tools from statistical information theory with possible applications of practical interest to any noisy imaging domains.

TUNING THE NOISE IN MAGNETIC RESONANCE IMAGING TO MAXIMIZE NONLINEAR INFORMATION TRANSMISSION

We demonstrate a new instance of useful-noise effect or stochastic resonance, occurring in magnetic resonance imaging (MRI). Based on the physics of signal–noise coupling specific to MRI, we establish the possibility of regimes where nonlinear post-processing can benefit from an increase in the level of the noise present in the MRI apparatus. The validation is obtained by both theoretical analysis and experimental observations. We especially show that the beneficial tuning of the noise can be practically achieved by controlling the bandwidth of the sampling receiver of the MRI apparatus. These results constitute a nontrivial extension of stochastic resonance in the domain of images, arising here with a signal–noise coupling in MRI which is distinct from the purely additive or multiplicative couplings previously investigated in the framework of useful-noise effect.

Constructive role of sensors nonlinearities in the acquisition of partially polarized speckle images

We study the impact of the level of the speckle noise on data acquisition in a partially polarized coherent imaging system with the presence of a nonlinearity in the imaging sensor characteristic. In perfectly linear acquisition conditions, due to the essentially multiplicative action of the speckle, the image contrast is unchanged as the speckle noise level increases, and so it has no impact on the quality of the acquired images. On the contrary, in nonlinear conditions the acquisition is affected by the speckle noise level. However, this effect of the speckle is not always detrimental. We show that, in definite nonlinear conditions, there is usually an optimal level of the speckle noise that leads to a maximum quality of the acquired images. We theoretically analyze such nonlinear regimes with partially polarized speckled images. We specifically exhibit the existence of an optimal speckle noise level in the interesting case of images realized only by a depolarization contrast. Illustrations are given with a simple 1-bit hard limiter and binary images. Then, we propose and discuss as perspectives an experimental optical setup to confront theory and experiment.