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

Recovering the ego-motion of a moving camera within a static scene supports many applications in robotics and computer vision. The presented work is motivated by pre-vis applications in the film industry; specifically the ability to render digital assets into the scene during production in real-time. A low-cost commodity depth camera can be easily mounted on and calibrated to a high quality production cameras and used to extract changes in sensor pose from the induced motion of the rigid scene. This work explores the effectiveness of the computationally efficient range flow technique to generate this real time pose information directly from the depth stream of a Kinect sensor. A number of challenges within the approach are addressed. First, an iterative version of the small rotationsmotion estimator is developed to ensure the most accurate inter-frame estimates. Second, the substantial issue of drift is addressed the accumulated error between true and estimated sensor pose as motion estimates are temporally integrated. Anchor frames which enjoy significant overlap with subsequent frames are stored and used to provide additional temporal range flow constraint within the estimation process. Where there are loops in the data sequence, it is advantageous to select anchors from previously seen parts of the scene. Finally, in some scene configurations, there is insufficient constraint from the depth images. We exploit the availability of registered intensity images to further constrain the sensor motion using the optical flow framework. Analogous to optical flow, range flow is a per-pixel constraint on the 3D displacement of an imaged 3D point given its local spatio-temporal depth derivatives. These must be combined across a region or an image to provide sufficient constraint to extract 3D motion[1, 2].

Accurate and Computationally-inexpensive Recovery of Ego-Motion using Optical Flow and Range Flow with Extended Temporal Support