GPU-enabled real-time panoramic background subtraction

---

Tezin Türü: Yüksek Lisans

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Enformatik Enstitüsü, Modelleme ve Simülasyon Anabilim Dalı, Türkiye

Tezin Onay Tarihi: 2015

Öğrenci: SERDAR BÜYÜKSARAÇ

Eş Danışman: GÖZDE AKAR, ALPTEKİN TEMİZEL

Özet:

Extraction of foreground objects using a Pan-Tilt camera is a challenging task for various video surveillance applications. It requires several steps such as camera motion extraction, image registration, panorama generation and background subtraction. All these steps require significant computing power. While achieving this by using only Central Processing Unit (CPU) is a challenging task, it might be enabled by efficient parallelization of the algorithms to run on Graphics Processing Unit (GPU). In this thesis an adaptive panoramic background generation and foreground object detection algorithm is implemented on GPU/CPU to run in real-time.