Express Waybill Extraction on Parcel Images by Logo Matching
International Conference on Information Photonics(2016)
摘要
We propose an efficient method for automatically extracting express waybills from parcel images, which is challenging due to varied resolution of parcel images, arbitrary direction of waybills and different information filled by senders. To address these challenges, logo matching is employed to extract the waybills. We begin by extracting scale-invariant feature-transformation (SIFT) keypoints from both the reference logo image and a parcel image, and matching them subject to a consistent projective transformation (homography) by using random sample consensus (RANSAC). Once the homography matrix is computed, we extract the waybill of parcel image by mapping all pixels from a standard waybill image to the parcel image. Experimental results on test datasets demonstrate the effectiveness of the proposed method.
更多查看译文
关键词
Logo matching,express waybill extraction,SIFT
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要