Reconstruction of Objects Using Lineage

Abstract

We study the problem of object reconstruction based on lineage, using photographs as our driving application. In addition to standard forward reconstructions, our model allows inverse transformations, reconstructions that exploit properties (e.g., commutativity), and imperfect reconstructions. With these additions, our model provides many more options for recovering a lost object. However, to choose among many possibly imperfect reconstructions, we need to carefully account for the accompanying ``degradation.'' In this paper, we propose a model for measuring degradation and a set of composition rules that help us measure the quality of reconstructions. Given this model, we propose an efficient algorithm for finding reconstructions and illustrate how it strikes a balance between efficiency and the quality of the produced results.