METHOD AND SYSTEM FOR MULTI-VIEW OBJECT SEGMENTATION USING DEPTH IMAGE

摘要

The present invention relates to a method and a system for separating a multi-view object using a depth image. The method of the present invention comprises: a system setup step of obtaining depth images corresponding to each target image on the basis of the depth images after viewpoints of reference color images and the depth images are matched, and generating an inter-viewpoint corresponding region on the basis thereof; and an object separating step of generating a single image energy term on the basis of color information on each target image, additionally generating an inter-viewpoint energy term reflecting the difference between an inter-viewpoint color and depth information on the basis of the inter-viewpoint corresponding region, and then separating an object included in each target image through a Markov random field (MRF) energy optimization. According to the present invention, an object separating operation is performed while maintaining inter-viewpoint consistency.