A Novel High-Order Statistics Map Based Method for Depth Map Generation of Static Scene

Abstract

Considering the increasing need of the contents of 3D video, it is significant to do some research in 2D-to-3D conversion. The depth map is necessary for this processing. In this paper, we propose an effective method to estimate the depth information of the low depth field images of static scenes. Firstly, construct a high-order statistics (HOS) map, which represents the high-frequency components of the input image. Then, conduct image segmentation to partition the image into different regions. Finally, assign the depth information to the regions and get a depth map.