No points

No Points tells the story of the logical bundle adjustment published in
IJITEE in 2020 with Mohamed Ben Ellefi: orienting spheres and cameras
without extracting a single point, by reasoning on occlusions alone.

Four chapters, to be handled more than read.

1. Seeing occlusion. Two discs on an image. Drag one and their relation is
written in James Allen's algebra, the thirteen ways two intervals can meet.
No coordinate is recorded; only the relation matters, and it reads with the
naked eye.

2. From interval to compass. Watch "a bit on the right" become
"north-east". North is the camera's optical axis, and every camera carries
its own compass rose.

3. The scene. The paper's own scene, six spheres and eight cameras, replayed
step by step: the machine orients the views by successive consistencies,
and touching an image shows the facts it contributed.

4. The reasoning machine. A real SPARQL engine, Oxigraph compiled to
WebAssembly, is embedded in the application and started on a tap. It loads
the Arpenteur ontology, finds the subclasses of Camera, then runs the
paper's very ASK queries in a loop, printing each conflict it resolves,
until the orientations match the native engine's. Locally, on your phone,
with no network of any kind.

The scope is stated rather than hidden: identical spheres on a single
plane, a discrete checkerboard space, cameras facing the scene. These are
the limits of a proof of concept, and the paper's future work lifts them
one by one.

Four languages, no account, no advertising, no data collection, and no
network permission at all: it works in airplane mode, and you can check
that yourself.

Based on M. Ben Ellefi and P. Drap, "A Bit on the Right, A Bit on the Left:
Towards Logical Bundle Adjustment", IJITEE 10(1), 2020,
doi 10.35940/ijitee.L8017.1110120. Engine: Oxigraph, Apache-2.0 / MIT.
Ontology: Arpenteur, image4d.fr/arpenteur.

IMAGE4D, image4d.fr