Journal of Pattern Recognition Research

Three-dimensional Flash Lidar camera is a new technology which allows a single camera to take multiple images in a fast succession. It produces a series of range delimited imagery using a single laser flash.  Three-dimensional Flash Lidar images capture images similar to the two-dimensional cameras, with the additional information of distance from the camera. The imagery can distinguish between objects at different distances, such as the sea-surface and underwater objects.   In this paper, we propose an algorithm to automatically find objects in 3D Flash Lidar images using a variational model of unsupervised multiphase image segmentation.  We apply a transformation to reduce the computational complexity; then, by taking a few coefficients, this reduced imagery (a scalar or vector-valued image) is used for object identification. Numerical results are presented to validate the approach both using scalar and vectorial models of multiphase segmentation.  This segmentation approach reduces the complexity of the problem and clearly distinguishes objects in the imagery.