Illumination-Aware Local Texture and Edge Preservation for Visible-Infrared Image Fusion Aydinlik Yönelimli Lokal Doku ve Kenar Korunumu ile Görünür-Infrared Görüntü Füzyon Analizi
34th Signal Processing and Communications Applications Conference, SIU 2026, İstanbul, Türkiye, 7 - 10 Temmuz 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/siu71813.2026.11636563
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: Infrared-RGB image fusion, quality metric, texture and edge analysis
- Kocaeli Üniversitesi Adresli: Evet
Özet
This study proposes a luminance-guided texture edge analysis for quality assessment in visible (RGB) and infrared (IR) image fusion. Existing metrics mostly rely on global evaluation and fail to capture local detail preservation. The proposed method utilizes a luminance map to perform locally weighted gradient-based similarity analysis. In addition, the metric is incorporated as a loss function component and evaluated on SwinFusion and TCDGNet models. Experimental results on LLVIP and RoadScene datasets demonstrate that the proposed approach provides more discriminative evaluation and improves fusion performance.