Implementation of Phase Based Optic Flow Algorithm for Obstacle Detection D'Souza Kevan N.1, Kadambi Govind R.2, Mohan Saima3 1M. Sc. [Engg.] Student, Department of Electronics and Computer Engineering, Bio Medical Engineering Centre, M.S. Ramaiah School of Advanced Studies, Bangalore, 560 054 2Professor and Head, Department of Electronics and Computer Engineering, Bio Medical Engineering Centre, M.S. Ramaiah School of Advanced Studies, Bangalore, 560 054 3Senior Lecturer, Bio Medical Engineering Centre, M.S. Ramaiah School of Advanced Studies, Bangalore, 560 054 Online published on 18 February, 2020. Abstract Optic flow is the apparent visual motion that one experiences during motion. A fundamental problem in Machine Vision and processing of image sequences is the estimation of the optic flow, which is the projection of the 3D surface point motion on a 2D sensor plane. This paper deals with the phase based algorithm for optic flow computation. The phase based optic flow technique is robust with respect to smooth shading and lighting variations and is amplitude invariant. Therefore, in this method phase contours are tracked over time. The image sequence is spatially filtered using a bank of Gabor filter quadrature pairs and its temporal phase gradient is computed. The velocity estimates with directions orthogonal to the filter pair's orientations are combined at a specific spatial location. The component velocity is rejected if the corresponding filter pair's information is not linear over time. A detailed study was carried out on the optic flow algorithms with a specific interest for its utility in Micro Air Vehicles (MAV). The related research by Temujin Gautama and Marc M. Van Hulle was selected and reviewed in view of its relevance to phase stability with respect to geometric deformations and its linearity as a function of spatial position. The selected algorithm was implemented through a developed MATLAB code and its results were correlated with that derived through a developed C code. The simulated results were also compared with pertinent-published results. The error metrics were identified and a performance evaluation of implemented phase based optic flow algorithm was carried out with the available standard synthetic image sequences. The implemented algorithm was also tested for its validity with the real time sequences. Top Keywords Machine Vision, Optic Flow, Phase Contours, Velocity Tuned Spatial Filters. Top |