Brussels Archives Accelerate Duplicate Image Purge as Digital Collection Hits New Milestone

Brussels city archivists confirmed this week that a coordinated duplicate-image removal operation is now running across at least four major public institutions, following a joint audit completed on June 30 that found an estimated 340,000 redundant image files clogging shared servers maintained by the Brussels Capital Region's digital heritage network.

The cleanup matters because the region is midway through a three-year digitisation push, and storage costs are real. Server space for the Regional Archives of Brussels-Capital at the Rue des Quatre Vents facility runs to roughly €180,000 per year in licensing and maintenance fees, according to budget documentation from the 2025 regional accounts. Duplicates — sometimes five or six copies of the same photograph generated by different scanning workflows — are inflating those costs and slowing search tools used by researchers and civil servants alike.

The timing is not accidental. The Brussels Urban Development Institute, working out of its Place Sainctelette offices, triggered the wider review in May when technicians migrating historic planning photographs to a new cloud platform discovered that a single block of Molenbeek street images from the 1970s had been scanned independently by three separate departments, producing 1,200 near-identical files. That discovery prompted the June 30 audit covering the Regional Archives, the Urban Development Institute, the Museum of the City of Brussels at the Grand Place, and the BOSA federal digitisation unit whose Brussels office sits on the Rue de la Loi.

What the Audit Found — and What Gets Deleted

The 340,000-file figure is a floor, not a ceiling. Auditors used perceptual hashing software to flag images where pixel similarity exceeded 98 percent, but they deliberately set the threshold high to avoid deleting legitimately distinct photographs that happen to show the same subject. Images scoring between 90 and 97 percent similarity — an additional batch of roughly 120,000 files — are being reviewed manually by a team of six archivists before any deletion decision is made.

Policy matters here. Under the Brussels Capital Region's Open Data Ordinance of 2021, any image already published on the open.brussels.be portal cannot simply be removed without a redirect record being created; broken URLs in third-party academic databases are a known problem from an earlier, messier purge the Regional Archives ran in 2019. This time, archivists are generating persistent identifier links for every deleted file, a process that adds approximately two weeks to the workflow but protects downstream users — university researchers, urban historians, journalists — who have embedded those image links in published work.

The Museum of the City of Brussels is taking a slightly different approach. Its digitisation manager has opted to merge duplicate records within the museum's own Axiell collections management system rather than outright deletion, preserving full provenance metadata for each scan even when the images themselves collapse into a single master file. The museum holds around 85,000 digitised photographs of Brussels street life dating back to the 1860s, and the deduplication exercise is expected to reduce active storage load there by around 18 percent once complete.

What Happens Next for the Public

Researchers using the iArchives public portal — the federated search interface that pulls from the Regional Archives and several municipal collections — should expect intermittent result gaps during the week of July 7 to 11, when the bulk deletion and redirect-linking work is scheduled to run. The BOSA unit has posted a service notice on its website advising users to download any specific images they need before July 7 if they require uninterrupted access.

Longer term, the institutions are pushing for a shared ingest protocol that would flag potential duplicates at the point of scanning rather than years later. A working group involving the four institutions held its first meeting at the Urban Development Institute on July 2 and is expected to produce a draft protocol by September. If adopted region-wide, it would apply to all new digitisation contracts tendered after January 1, 2027.

For the average Brussels resident, the practical upshot is a cleaner, faster image search for historical neighbourhood photographs — the Ixelles pond, the old Anneessens gate, the pre-renovation Rue Neuve — with fewer dead-end duplicate results cluttering what should be a straightforward digital library.