Rounding noise effects' reduction for estimated movement of speckle patterns

The problem of resolution enhancement for speckle patterns analysis-based movement estimation is considered. In our previous publications we showed that this movement represents the corresponding tilt vibrations of the illuminated object and can be measured as a relative spatial shift between time adjacent images of the speckle pattern. In this paper we show how to overcome the resolution limitation obtained when using an optical sensor available in an optical mouse and which measures the Cartesian coordinates of the shift as an integer number of pixels. To overcome such a resolution limitation, it is proposed here to use simultaneous measurements from the same illuminated spot by a few cameras (sensors) each having imaging lenses with different amounts of defocusing. The amount of defocusing defines the proportion ratio between actual changes in the tilt plane and measured shift between speckle images. To utilize the diversity of such ratios we apply a beam-forming signal processing approach that makes it possible to achieve different design criteria and improve the measurement accuracy, respectively. The validity and properties of the proposed solution are demonstrated by a few examples of in-vivo touchless measurements of human heart beat sounds.