Digital Holography is a strong imaging method for 3D vision and display systems. But, the use of coherent light sources reveals annoying visual procedures, namely speckle noise, an inherent interference effect because of the laser. Such sort of coherent noise, superposing to digital holograms, majorly degrades the corresponding re-building quality in holographic systems.
The diminishing of the light coherence, by structuring the laser source or by linking and recording multiple holograms, were the two major identified methods to address such problem. In specific, multi-look Digital holography has proved as one of the most effective methods to enhance the quality of both the optical and numerical reconstructions. Nevertheless, numerous methods have been introduced to diminish holograms noise by implementing tailor-made procedures that are typically based on numerical reconstructions of digital holograms for visualization of image only.
Indeed, just few of the methods, functioning directly on recorded holograms have been enclosed, ensuring the enhancement of the quality for optical display procedures. Among such methods, the 3D Block Matching filtering or BM3D has illustrated very robust de-noising potentials in the arena of digital image processing, by means of a block collaborative and grouping filtering strategy.
But, this technique needs a specific level of the initial signal – to – noise ratio or SNR of images to be processed; otherwise, an inaccurate grouping could occur, reducing down the reconstruction quality. To combat this challenge, a preliminary filtering is mainly employed in the case of images with low SNR, as in the case of digital holograms. We identified that the joint action of MLDH, collaborative and grouping filtering enables to accomplish high-quality numerical reconstructions in digital holography.
We consider to this proposed technique as MLDH – BM3D. In specific, MLDH pre-processing permits to accomplish the enhanced grouping step allowing better working conditions for the iterative processing blocks of the associative sparse 3D filtering.
Definitely, we illustrated that BM3D and MLDH can be considered as complementary steps, mixing smart optical recording techniques and numerical processing. In the labs at Institute of Applied Sciences and Intelligent Systems of CNR, we illustrated that the identified approach functions efficiently for both multi-wave and single-wave lengths digital holography, by accomplishing an enhancement up to 98% of noise suppression, thus illustrating an unmatched quality in holographic 3D reconstructions that can be considered ‘noise-free’ for humans.
These results are highly impressive and lead to pave the way to the next further generations of holographic imaging systems that are based on laser technology. Now it is to be seen that how much this research would benefit further studies.
Filed Under: News