PriOr-Flow: Enhancing Primitive Panoramic Optical Flow with Orthogonal View

2025-06-30

AI Summary

Panoramic optical flow estimation suffers from severe distortion near the poles due to spherical projection (e.g., equirectangular projection), drastically degrading the performance of conventional perspective-based optical flow methods. To address this, we propose a dual-branch framework that jointly leverages the original panoramic view and a low-distortion orthographic view—achieving, for the first time, cost-volume-level fusion between them. Our key innovations include: (i) the Dual-Cost Collaborative Lookup (DCCL) operator, enabling precise cross-view cost-volume alignment and fusion; and (ii) the Orthographic-Driven Distortion Compensation (ODDC) module, which explicitly suppresses distortion-induced noise in feature space. The method is fully compatible with mainstream perspective optical flow architectures and supports iterative refinement. Evaluated on standard panoramic optical flow benchmarks, it achieves state-of-the-art performance, with significant error reduction near the poles—establishing a new benchmark for wide-field-of-view motion estimation.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Distributed Search
• Intelligent Robots: State Estimation

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Data management and stream processing for Web, mobile and wireless applications
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Panoramic optical flow enables a comprehensive understanding of temporal dynamics across wide fields of view. However, severe distortions caused by sphere-to-plane projections, such as the equirectangular projection (ERP), significantly degrade the performance of conventional perspective-based optical flow methods, especially in polar regions. To address this challenge, we propose PriOr-Flow, a novel dual-branch framework that leverages the low-distortion nature of the orthogonal view to enhance optical flow estimation in these regions. Specifically, we introduce the Dual-Cost Collaborative Lookup (DCCL) operator, which jointly retrieves correlation information from both the primitive and orthogonal cost volumes, effectively mitigating distortion noise during cost volume construction. Furthermore, our Ortho-Driven Distortion Compensation (ODDC) module iteratively refines motion features from both branches, further suppressing polar distortions. Extensive experiments demonstrate that PriOr-Flow is compatible with various perspective-based iterative optical flow methods and consistently achieves state-of-the-art performance on publicly available panoramic optical flow datasets, setting a new benchmark for wide-field motion estimation. The code is publicly available at: https://github.com/longliangLiu/PriOr-Flow.

Problem

• Addressing distortion in panoramic optical flow estimation
• Enhancing optical flow accuracy in polar regions
• Mitigating noise in sphere-to-plane projection distortions

Innovation

• Dual-branch framework enhances panoramic optical flow
• DCCL operator mitigates distortion noise effectively
• ODDC module refines motion features iteratively