On k-full-view-coverage-algorithms in camera sensor networks

Many security and reliability applications require guaranteed k-coverage of an area at all times. As a result, k-coverage has attracted much attention in the past decade, but most of the previous work on k-coverage has focused on traditional scalar sensors instead of camera sensors. Camera sensors qualify as vector sensors because cameras from different positions can form various views of the target. The sensing quality of camera sensors depends on the angle between the facing direction of the target and the viewing direction of the camera. The concept of full-view coverage [1] is of great significance because it ensures both the detection and recognition of a target. A target experiences full-view coverage if, regardless of the direction it faces, it is always in view of a camera. At the same time, the camera's viewing direction should be sufficiently close to the direction the target faces [1]. This research intends to investigate a special case of k-coverage called k-full-view-coverage. This case is where the target is in an over-deployed camera sensor network and k disjoint subsets of cameras are able to provide full-view coverage of the target. The two proposed algorithms aim to k-full-view cover both a single target and a given area based on equilateral triangle displacement.