Full GPGPU implementation of 2D-3D conversion on laptop GPU

This paper describes the General-Purpose computing on Graphics Processing Unit (GPGPU) optimized implementation of 2D-3D Conversion. 2D-3D Conversion is large-scale image processing and consists of 170 kernels. Those kernels, which are tasks performed by GPU, contain various algorithms. 30fps real-time processing of 2D-3D Conversion conventionally requires special hardware since it is computationally expensive. To clarify the feasibility of real-time processing by software, we did full GPGPU implementation of all image processing of 2D-3D Conversion on a laptop PC.