In this section is to give a brief description of dynamic programming algorithm that produces an optimal encoding under on very reasonable assumption. A more detailed description can be found in Motta al.. Although this is not likely to be precisely true in practice, it is likely to be a very good approximation to what happens. That is, two encodings of a frame at the same quality are likely to be equivalent in their ability to predict a subsequent frame.
The dynamic sub window scheme is summarized as follows: If 1st frame eq kth then Skip the transcoding of the k+1th sub-window else Transcode the frame no=2
Hyper spectral images contain a wealth of data, but hyper spectral images are interpreting them requires an understanding of exactly what properties of ground materials we are references is provided on trying to measure, and how they relate to the measurements actually made by the hyper spectral sensor.
In multi spectral imaging, a series of images acquired at many wavelength producing an “image cube” in Figure 1.
Figure 1 : Image cube
DYNAMIC HYPER SPECTRAL VIDEO FRAME SKIPPING
For the purposes of this paper, the planes of decreasing spectral energy were numbered beginning with the lowest level sub-band plane. Then, the multiplier for each of the data points in each plane was determined as follows
To assume that at a time t=n a new video frame should be transmitted through the variable network and at this time t the available bandwidth, say B(n), is less than the minimum required for transmitting a frame in one frame period, In case that the current bandwidth is greater than the requested one, all multimedia information can be delivered. A better solution is proposed here that performs a content-based sampling. In this way, among the current and the candidate frames for skipping (i.e., k, k+1, k+2, …,k+K-1), the most representative is selected to be delivered whereas the remaining frames are discarded.
Figure 2 : Video frame skipping
We calculate the average feature vector f over all K frames F=1/K XVif
To select as most representative frame for the one whose feature vector is closer in the sense of the L2 norm. Figure 3 presents a graphical representation of the proposed scheme. In this example, at time t=n, the algorithm estimates that the bandwidth is four times lower than the minimum required, i.e., K=4. Then, the algorithm selects as representative among the four successive frames the one whose feature vector is the closest to the min vector of the four frames. In this example, we assume that this frame is the second, i.e., J=2. We have also assumed that this frame completes its transmission within three frames instead of four, due to the fact the network bandwidth has slightly increase during the frame transmission. Thus, at time n=4 the current bandwidth is again evaluated resulting in K=5.