다중 시점 영상을 이용한 파노라마 영상 생성 및 물체 인식 알고리즘

Multi-view images taken at different viewpoints provide us rich information about the scene with additional field-of-view(FOV) and three-dimensional information. In this paper, we introduce new methods for image stitching and object recognition using multi-view images. To achieve seamless image stiching for dynamic scene, we propose a new color tone mapping method. In addition, we propose a graph-based object recognition method aiming at reducing ambiguity in recognition.