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

In this work, we propose a novel approach to prioritize the depth mapcomputation of multi-view stereo (MVS) to obtain compact 3D point clouds ofhigh quality and completeness at low computational cost. Our prioritizationapproach operates before the MVS algorithm is executed and consists of twosteps. In the first step, we aim to find a good set of matching partners foreach view. In the second step, we rank the resulting view clusters (i.e. keyviews with matching partners) according to their impact on the fulfillment ofdesired quality parameters such as completeness, ground resolution andaccuracy. Additional to geometric analysis, we use a novel machine learningtechnique for training a confidence predictor. The purpose of this confidencepredictor is to estimate the chances of a successful depth reconstruction foreach pixel in each image for one specific MVS algorithm based on the RGB imagesand the image constellation. The underlying machine learning technique does notrequire any ground truth or manually labeled data for training, but insteadadapts ideas from depth map fusion for providing a supervision signal. Thetrained confidence predictor allows us to evaluate the quality of imageconstellations and their potential impact to the resulting 3D reconstructionand thus builds a solid foundation for our prioritization approach. In ourexperiments, we are thus able to reach more than 70% of the maximal reachablequality fulfillment using only 5% of the available images as key views. Forevaluating our approach within and across different domains, we use twocompletely different scenarios, i.e. cultural heritage preservation andreconstruction of single family houses.

Prioritized Multi-View Stereo Depth Map Generation Using Confidence Prediction