DJAREDDIR ,KawtherTOUIL, GHASSEN2025-11-122025-11-122025http://dspace.univ-skikda.dz:4000/handle/123456789/5394This project explores video stitching—a process that merges multiple video streams into a single panoramic output. Building on image stitching techniques like feature detection, homography estimation, and blending, it addresses video-specific challenges such as temporal consistency, motion handling, and real-time processing. A Python-based application was developed using OpenCV and PyQt6, offering both simple and advanced stitching modes. Advanced methods include hybrid stitching, attention-based blending, and GAN-based inpainting. Evaluations show that while fast methods are efficient, advanced techniques produce superior visual quality, making the system suitable for applications in VR, surveillance, and mappingenMémoire de Master