Innovative local texture descriptor in joint of human-based color features for content-based image retrieval
Image retrieval is one of the hot research topics in computer vision which has been paid much attention by researchers in the last decade. Image retrieval refers to retrieving more similar images to the query form a huge image database. It is used widely in different scopes such as medical and searc...
Gespeichert in:
Veröffentlicht in: | Signal, image and video processing image and video processing, 2023-11, Vol.17 (8), p.4009-4017 |
---|---|
Hauptverfasser: | , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | Image retrieval is one of the hot research topics in computer vision which has been paid much attention by researchers in the last decade. Image retrieval refers to retrieving more similar images to the query form a huge image database. It is used widely in different scopes such as medical and search engines. Texture and color information play an important role in image content recognition. So, in this paper an innovative approach is proposed based on a combination of color and texture features. In this respect, an extended version of local neighborhood difference patterns (ELNDP) is proposed for the first time to achieve discriminative features. The ELNDP exploits the advantages of LBP and LNDP texture descriptors. Also, for global features extraction, optimized color histogram features in HSV color space are used to extract color features. Finally, the extended Canberra distance metric is used to retrieve more relevant images which is not sensitive to lower values like classic Canberra. The performance of the proposed approach is evaluated on five benchmark datasets such as Corel 1 K, 5 K, 10 K, STex and Colored Brodatz. The results are evaluated in terms of average precision rate (APR), average recall rate (ARR), The experimental results show that the proposed approach provides higher retrieval performance in comparison with state-of-the-art methods in this area such as machine learning-based and deep learning-based approaches. |
---|---|
ISSN: | 1863-1703 1863-1711 |
DOI: | 10.1007/s11760-023-02631-x |