SIMILARITY RATIO BASED ALGORITHMS TO GENERATE SAR SUPERPIXELS

---

Tezin Türü: Doktora

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Fen Bilimleri Enstitüsü, Fen Bilimleri Enstitüsü, Türkiye

Tezin Onay Tarihi: 2017

Tezin Dili: İngilizce

Öğrenci: EMRE AKYILMAZ

Asıl Danışman (Eş Danışmanlı Tezler İçin): Uğur Murat Leloğlu

Eş Danışman: İlkay Ulusoy

Özet:

Synthetic Aperture Radar (SAR) has the capability of working in all weather conditions during day and night that makes it attractive to be used for automatic target detection and recognition purposes. However, it has the problem of high amount of multiplicative speckle noise. Superpixel segmentation as a preprocessing step is an oversegmentation technique that groups similar neighboring pixels into regularly organized segments with approximately the same size. As boundaries of the objects are important elements to be traced, superpixels should adhere well to the edges. This can only be achieved by an algorithm robust to speckle noise. In this thesis, similarity ratio is first developed as a new metric that is robust to speckle noise. Secondly, Mahalanobis distance is used instead of Euclidian so that the superpixel can fit better to shapes in the real world. Thirdly, the constant determining the relative importance of radiometric and geometric terms is replaced with an adaptive function. The performance of combinations of similarity ratio with Euclidean distance (SREP), Mahalanobis distance (SRMP) and Mahalanobis distance with adaptive scheme (SRAMP) are evaluated by conducting experiments on real and synthetic images. The experimental results showed that similarity ratio and adaptive Mahalanobis proximity (SRAMP) outperforms the other approaches in terms of uniformity, compactness and visual appearance.