Real-Time 2D-to-3D Conversion for Television Broadcasting Based on Embedded System

In this paper, we present an efficient 3D display technique that employs current 2D TV channels. First, a gradient background depth is generated for the still region. To create a stereo effect, multiple objects are detected and segmented using the binary format. Object depth is estimated according to the object size and its location. Depth value is assigned as a proportion of the object size using the linear quantitative method. Each object size is calculated and its depth value is proportional to the size parameter. If an object is the largest one and is located in the central region, the maximum depth value is assigned. For real-time implementation, dual ARM cores perform the task using the algorithm of the proposed stereo conversion. Our results show that a low-cost embedded system can perform the proposed 3D algorithm, with a processing time of 43ms per frame. The signal from a TV channel can be sampled and converted to a 3D format display on a conventional LCD screen.