Hexagonal sampling

=== Aliasing effects minimized by the use of optimal sampling grids ===

=== Aliasing effects minimized by the use of optimal sampling grids ===

Recent advances in the [[charge-coupled device|CCD]] technology has made hexagonal sampling feasible for real life applications. Historically, because of technology constraints, detector arrays were implemented only on 2-dimensional rectangular sampling lattices with rectangular shape detectors. But the super [CCD] detector introduced by ”’Fuji”’ has an octagonal shaped pixel in a hexagonal grid. Theoretically, the performance of the detector was greatly increased by introducing an octagonal pixel. The number of pixels required to represent the sample was reduced and there was significant improvement in the [[signal-to-noise ratio|Signal-to-Noise Ratio]] (SNR) when compared with that of a rectangular pixel.<ref>{{Cite book | last1 = R. Vitulli | first1 = R. | last2 = Del Bello | first2 = U. | last3 = Armbruster | first3 = P. | last4 = Baronti | first4 = S. | last5 = Santurti | first5 = L. | chapter = Aliasing effects mitigation by optimised sampling grids and impact on image acquisition chains | doi = 10.1109/IGARSS.2002.1025749 | title = IEEE International Geoscience and Remote Sensing Symposium | volume = 2 | pages = 979 | year = 2002 | isbn = 0-7803-7536-X }}</ref> But the drawback of using hexagonal pixels is that the associated [[Microlens|fill factor]] will be less than 82%. An alternative method would be to interpolate hexagonal pixels in such a manner that we ultimately end up with a rectangular grid. The [[SPOT (satellite)|Spot]]&nbsp;5 [[satellite]] incorporates a similar technique where two identical linear CCD’s transmit two [[quasi]]-identical images that are shifted by half a pixel. On interpolating the two images and processing them, the functioning of a detector with a hexagonal pixel is mimicked.

Recent advances in the [[charge-coupled device|CCD]] technology has made hexagonal sampling feasible for real life applications. Historically, because of technology constraints, detector arrays were implemented only on 2-dimensional rectangular sampling lattices with rectangular shape detectors. But the super [CCD] detector introduced by ”’Fuji”’ has an octagonal shaped pixel in a hexagonal grid. Theoretically, the performance of the detector was greatly increased by introducing an octagonal pixel. The number of pixels required to represent the sample was reduced and there was significant improvement in the [[signal-to-noise ratio|Signal-to-Noise Ratio]] (SNR) when compared with that of a rectangular pixel.<ref>{{Cite book | last1 = R. Vitulli | first1 = R. | last2 = Del Bello | first2 = U. | last3 = Armbruster | first3 = P. | last4 = Baronti | first4 = S. | last5 = Santurti | first5 = L. | chapter = Aliasing effects mitigation by optimised sampling grids and impact on image acquisition chains | doi = 10.1109/IGARSS.2002.1025749 | title = IEEE International Geoscience and Remote Sensing Symposium | volume = 2 | pages = 979 | year = 2002 | isbn = 0-7803-7536-X }}</ref> But the drawback of using hexagonal pixels is that the associated [[Microlens|fill factor]] will be less than 82%. An alternative method would be to interpolate hexagonal pixels in such a manner that we ultimately end up with a rectangular grid. The [[SPOT (satellite)|Spot]]&nbsp;5 [[satellite]] incorporates a similar technique where two identical linear CCD’s transmit two quasi-identical images that are shifted by half a pixel. On interpolating the two images and processing them, the functioning of a detector with a hexagonal pixel is mimicked.

=== Hexagonal structure for Intelligent vision ===

=== Hexagonal structure for Intelligent vision ===