A Framework Enabling Object-Based Queries Within Images on Digital Documents
IEEE Access, cilt.14, ss.83316-83323, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 14
- Basım Tarihi: 2026
- Doi Numarası: 10.1109/access.2026.3698648
- Dergi Adı: IEEE Access
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, INSPEC, Directory of Open Access Journals
- Sayfa Sayıları: ss.83316-83323
- Anahtar Kelimeler: digital archives, document image analysis, document layout analysis, JSON-based metadata, logical query processing, object detection, Object-based document search, visual document retrieval, visual metadata indexing
- Kocaeli Üniversitesi Adresli: Evet
Özet
Digital documents frequently contain visual information such as photographs, illustrations, charts, and graphical elements that cannot be accessed through conventional text-based search mechanisms. This study proposes an object-based visual querying framework for digital documents. The proposed framework first applies document layout analysis to detect image regions within PDF pages, then performs object and chart detection on the extracted images. The detected classes, confidence scores, page numbers, image identifiers, and bounding-box coordinates are stored in a hierarchical JSON-based metadata structure. A logical query mechanism is then used to retrieve visual content according to object-based expressions containing AND, OR, and NOT operators. Finally, the matching objects are highlighted on the original document pages and a reconstructed result document is generated. Experimental results show that the proposed framework enables object-level search over visual document content and provides a scalable basis for visual indexing in digital archives. The system is particularly useful for large document collections in which manual inspection of images is impractical.