Double-Helix Vision (DH-V2): A Geometry-Based Visual Sampler for Bandwidth-Constrained Perception

A new geometry-based visual sampler compresses 4K images by a factor of 1,433 without neural networks, according to a preprint posted to arXiv on June 9, 2026 [1][2]. The method, called Double-Helix Vision (DH-V2), converts 2D frames into compact 1D signals using paired spiral trajectories and runs its full perception pipeline in under one millisecond on CPU-only hardware [1]. The sampler uses two phase-shifted helices, designated Alpha and Beta and offset by 180 degrees, to scan an image with a foveation pattern that concentrates sampling points at the center and reduces them toward the periphery [1]. At 4K resolution, the approach achieves a 1,433x compression ratio, equivalent to a 99.93% reduction in data volume, while retaining the geometric structure of the scene [1][2]. The full pipeline — which includes spatial mapping, temporal collision detection, and intra-frame structural disparity estimation — completes in 0.52 ms at 1080p on CPU-only hardware, with no neural network dependencies [1]. On the CIFAR-10 benchmark, DH-V2 recorded a +6.03% accuracy gain over uniform random sampling when operating at an extreme sampling budget of K=128 points per helix [1][2]. The authors also provide a JSON-serializable Robotics API that delivers sub-millisecond spatial perception reports in 2.7 KB packets [1]. Code and benchmarks are released under the MIT License [1]. The paper appears on arXiv, an open-access repository that hosts electronic preprints across physics, mathematics, computer science, and related fields [6]. arXiv was founded in 1991 and, as of November 2024, receives approximately 24,000 submissions per month [6]. The repository surpassed two million articles by the end of 2021 [6]. Submissions are moderated but not peer-reviewed, a status shared by the DH-V2 preprint [1][6]. The article page also surfaces several community-built discovery tools through the arXivLabs framework, which allows third-party collaborators to integrate experimental features directly on the site [4][5]. Among the tools linked to the DH-V2 abstract are the Bibliographic Explorer for citation-tree navigation, Connected Papers, and Litmaps for interactive literature mapping [5]. arXivLabs projects operate under guidelines that require partners to uphold values of openness, community, and user data privacy, and collaborators receive only minimal, anonymized user data necessary for feature functionality [4].