Image sequence correspondence via Hopfield neural network
Mong‐Fong Horng, Jia‐Yaw Chang, S.W. Lee, W.S. Hsu, Monica Cheng · 1991
Summary form only given. A neural network approach for finding trajectories of feature points in a monocular image sequence is proposed. The Hopfield neural network has been applied to the image sequence correspondence. Example and simulation results were obtained to illustrate the design process and the convergence characteristics of the proposed neural network. Through the massively parallel processing power of the neural network, a real-time and accurate solution can be obtained.>