Image Transformation of Wide-Angle Ceiling Camera for People Detection

DOI

Bibliographic Information

Other Title
  • 人物検出のための広角天井カメラ画像の変換法

Abstract

In recent years, many systems have been developed to detect people from surveillance camera images. Especially, wide-angle ceiling cameras are widely used because they can cover a wide surveillance area by a single camera. However, the camera height change causes non-linear image deformation, and the detection performance is significantly degraded. In this study, we propose an image transformation method that simulates the height change of a wide-angle ceiling camera. We also propose a method that estimates image transformation parameters by Bayesian optimization. We confirmed the detection accuracy is improved for the images captured at different height by applying our method.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top