More specifically, the team proposed a Keller cone-based imaging projection kernel. This kernel is implicitly a function of edge orientation, and this relationship is exploited to infer edge existence/orientation through hypothesis testing over a small set of possible edge orientations. In other words, if the presence of an edge is determined, the edge orientation is selected that best matches the resulting Keller cone-based signature for a particular point of interest in imaging.
“There are local dependencies at the edges of real objects,” said lead researcher Anurag Paraprol. project students. “Therefore, once we find a reliable edge point via the proposed imaging kernel, we use Bayesian information propagation to propagate that information to the remaining points. It can further help you improve your images. can It will be overwhelmed by other edges closer to the transmitter. ” Finally, once the image is formed, researchers can use image completion tools from the visual domain to further improve the image.
“It is worth noting that conventional imaging techniques suffer from poor imaging quality when deployed in commercially available WiFi transceivers. Because there will be none left,” Pallaprolu added.
The researchers also extensively studied the effects of several different parameters, including surface curvature, edge orientation, distance to the receiver grid, and transmitter location, on the Keller cone and the proposed edge-based imaging system. and thereby a systematic imaging system design.