Cutset Image Sampling for Reconstruction, Compression and Sensor Networks

Cutset image sampling is a new approach to image sampling in which the image is sampled densely along a grid of lines, e.g. a Manhattan grid, instead of the conventional uniformly spaced samples, as in rectangular or hexagonal sampling. It is useful in situations where edge image information must be preserved, and where samples are taken by a vehicle or by a wired or wireless sensor network. In the latter cases, the closer spacing of adjacent sensors in cutset sampling requires less wire or energy than conventional sampling for intersensor communication. It is also useful as a first step in image compression, where the closer spacing of samples increases their correlation and compressibility. This talk introduces cutset sampling, as well as image reconstruction methods and a sampling theorem for cutset samples, lossless and lossy image compression methods based on cutset sampling, and source localization with a network of sensors deployed in a Manhattan grid.