The aim of this study is to develop a method for locating objects in conditions of their partial visibility and possible surface shape changes, with a focus on parts manufactured using selective laser sintering (SLS). This paper examines methods for achieving visibility of a hidden part in a video image and proposes new approaches to achieving this goal, one cautious and one confidence. Particular attention is paid to the architecture of a neural network designed to identify parts whose shape has been noticeably altered due to powder adhering to the surface. A general framework for solving the problem of identifying and retrieving an object from a sintering mold is also presented, and the role of an integrated object identification system in the overall technological solution is determined. The practical significance of this study lies in the development of a method for achieving visibility, detection, and identification of parts located in a powder environment, which is relevant for the automation of post-processing processes in industries using selective laser sintering technology.
Chernyshov et al. (Thu,) studied this question.