키포인트 기반의 시각적 지역 특징을 이용한 평면 물체 인식: 비교 분석

초록

This paper presents a comparative analysis of what factors should be considered when recognizing a planar object using keypoint-based visual local features. Using a database that considers various scenarios for planar object recognition that may appear in real-world applications, we compare and analyze the recognition accuracy and overall computation time while differently from the local feature extraction algorithm, matching and refinement algorithm, and homography calculation algorithm. The comparative analysis presented in this paper can be used as a good reference when developing computer vision or robot vision applications that need to recognize planar objects using keypoint-based visual local features. In particular, it is expected to be useful when developing applications that require estimating the three-dimensional posture of a camera based on matching local features and calculating homography, such as augmented reality or simultaneous localization and mapping.

키워드