13.8 Real Image Experiments

In this section, real hyperspectral image experiments were conducted based on the 15-panel HYDICE image scene in Figure 13.2 (see Figure 1.15(a)). One major difference between the real HYDICE image scene in Figure 13.2 and the simulated synthetic image in Figure 13.1(c) is that there is very little knowledge of the image background in Figure 13.2 compared to the image background in Figure 13.1(b) that was simulated by complete knowledge. As expected, FLSMA may not perform as well as it did for the synthetic image if the image background in Figure 13.1(b) was not well characterized. In order to demonstrate this fact, two scenarios are used to characterize the image background as follows. Additionally, we also assume that the knowledge of the 9 R pixels in the img and 5 R pixels in img panels in Figure 1.15(b) is available a priori. So, the panel signatures in Figure 13.2 and 14 R pixels in both the img and img panels are considered to be prior knowledge according to the ground truth in Figure 1.15(b).

13.8.1 Image Background Characterized by Supervised Knowledge

By viewing ...

Get Hyperspectral Data Processing: Algorithm Design and Analysis now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.