2007/06/01 by Yongwei Sheng
Computer Science · Engineering · #Advanced Vision and Imaging #Image Processing Techniques and Applications #Robotics and Sensor-Based Localization
paper · doi:10.1111/j.1477-9730.2007.00425.x
crossref issued 2007/06/01 · crossref published 2007/06/01 · crossref published-print 2007/06/01 · openalex publication_date 2007/06/01 · crossref published-online 2007/08/06 · crossref created 2007/08/07 · crossref deposited 2023/10/14 · openalex created_date 2025/10/10 · crossref indexed 2026/07/28 · openalex updated_date 2026/07/29
Abstract True ortho‐images are generated from perspective images with both object and terrain displacement distortion corrected and with occlusions removed. However, widely used algorithms based on a Z‐buffer have drawbacks and may result in artefacts in the generated ortho‐images. This paper analyses the artefacts induced by Z‐buffer algorithms in true ortho‐image generation, and proposes a vector‐based method to avoid algorithm‐induced artefacts. The proposed method treats a perspective image pixel as a square patch instead of an image pixel point, and generates ortho‐images in the vector domain. Under the vector implementation, occlusion detection is a simple process, and image resampling becomes a straightforward procedure. Most importantly, this method can minimise artefacts in the ortho‐images so generated.