LiDAR & Satellite Incorporated Application(59)
A Studied Case
Overview (2)
A Problem Existing in Dealing with Hyper-Data (2)
Figure 3
Figure 4
According to information in head file for bad merged imagery (Figure 1) from Figure 3, ones tried to select some bands such as VIS, VINR from multi-band georeferenced imageries to mix them with collected hyper-data so that the hyper-data are thereby referenced. The idea seems very good, but they may not realize a few important points:
1. The hyper-data are in fact stored into a multi-dimensional matrix. Mathematically speaking, even if spatial points at two-dimensional plain are exactly same, but a big problem appears in the third dimension (see Figure 4). Ten bands (see Figure 3) are linearly added into 124 bands, vice versa. A fact is that the curve appears at each band of hyper-spectrum is a continuous function (usefully, an unknown function in remote sensing) having corresponding FWHM (Full Width at Half Maximum). FWHM is an expression of the extent of a function given by the difference between the two extreme values of the independent variable at which the dependent variable is equal to half of its maximum value. Based on this vital relation, values of FWHM are offered in ENVI so that the known values (intensities: radiance and reflectance) acquired by measurement from survey against the corresponding wavelengths are confirmed. Hence, a curve is then formed. A quick example for easily understanding is given here: it seems that a drop of ink is added into a large-volume container of water if do so.
2. Some useful and important concepts of relevant difference should be pointed out as follows:
• The measured radiance at radiance is a series of points (thus sufficient points) to form a continuous function because it is indirectly generated (reflected) by solar beam (photon flux). Accordingly, the curve must be a continuous function with respect to wavelength or wavenumber in math by means of FWHM.
• The data stored in multi-band georeferenced imageries such as digital number (DN) are discontinuous points, which fail to form a continuous function against wavelength or wavenumber. It can form a sectional function (see Figure 2).
3. Very clear, all of hyper-data are damaged if the treatment of hyper-spectra is performed in such a manner. Therefore, strongly suggest ones to quit treating them in this manner, replaced by another.
The other manner of solving this problem is to be introduced later. Hope readers to give feedback and comments on them for discussion and improvement.
