Current Proceedings on Technology

Yazarlar: Muhammet Talha Büyükakkaşlar, Nurettin Şenyer, Recai Oktas

Konular:-

Anahtar Kelimeler:Gait analysis,Shadow biometric,Shadow extraction,Hough transform

Özet: This study explores implementation of a “Automated Shadow Separator” which separates humans body and shadow from foreground image of body and its shadow. SDHA2010 Areal View Challenge dataset is used for this study. Gait Analyses based on areal long range videos fails because of lack of body information. Constructing gait classifier/analysis based on shadow is a new idea proposed by Stoica. First step of this process is separating body from shadow. In this study, body shadow separation point is calculated in two stage. First, two lines calculated by Hough Transform to represent body and shadow. After that intersection point of these lines defined as body shadow separation point. BW skeleton is extracted from human and shadow containing image for the method developed. Proposed method uses Hough Transform which is more robust at extracting characteristics of uncertain images. With this method two most voted pairs are selected and intersection of these points’ represent lines is considered as separation point. Ground-truth data calculated support of users is compared with Hough Transform method and Stoica’ separation algorithm. We will use shadow for gait analysis.

---