English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Existing volumetric neural rendering techniques, such as Neural RadianceFields (NeRF), face limitations in synthesizing high-quality novel views whenthe camera poses of input images are imperfect. To address this issue, wepropose a novel 3D reconstruction framework that enables simultaneousoptimization of camera poses, dubbed CBARF (Cascaded Bundle-Adjusting NeRF).Ina nutshell, our framework optimizes camera poses in a coarse-to-fine manner andthen reconstructs scenes based on the rectified poses. It is observed that theinitialization of camera poses has a significant impact on the performance ofbundle-adjustment (BA). Therefore, we cascade multiple BA modules at differentscales to progressively improve the camera poses. Meanwhile, we develop aneighbor-replacement strategy to further optimize the results of BA in eachstage. In this step, we introduce a novel criterion to effectively identifypoorly estimated camera poses. Then we replace them with the poses ofneighboring cameras, thus further eliminating the impact of inaccurate cameraposes. Once camera poses have been optimized, we employ a density voxel grid togenerate high-quality 3D reconstructed scenes and images in novel views.Experimental results demonstrate that our CBARF model achieves state-of-the-artperformance in both pose optimization and novel view synthesis, especially inthe existence of large camera pose noise.

CBARF: Cascaded Bundle-Adjusting Neural Radiance Fields from Imperfect Camera Poses