An Automatic Compensation System for Unclear Area in 360-degree Images Using Pan-tilt Camera

摘要

We conduct research using a 360-degree camera THETA. The ability to capture a wider range with a single camera is very advantageous for object recognition and classification. However, processing of images taken with a 360degree camera has problems such as image distortion caused by the lens and insufficient resolution in recognizing distant objects, so that it is difficult to classify the object in the corresponding area. In this paper, we propose a method for modifying unclear areas of 360-degree camera images using images taken with a couple of Pan-Tilt (PT) cameras those are installed on a same platform as THEA. By using this system with three cameras, it can be expected that object classification in an unclear area, which has been difficult in the past, is performed with high accuracy.