Implementation of the Lucas-Kanade image registration algorithm on a GPU for 3D computational platform stabilisation

Abstract

Image registration forms the basis of many computer vision tasks. The Lucas-Kanade image registration algorithm is known to efficiently solve the sub-problem of rigid image registration. It is therefore often used in image stabilisation applications. This paper presents the details of a real-time implementation of the Lucas- Kanade image registration algorithm on a Graphics Processing Unit (GPU) using the OpenGL Shading Language (GLSL). The implementation is driven by a real world requirement to computationally stabilise the undulatory motion of an ocean-based wide area surveillance system.