Displacement vector estimation with cellular neural networks
D. Feiden, Ronald Tetzlaff · 2003
Displacement vector estimation is one of the open key problems in computer vision and video coding. For example, in computer vision, displacement vector estimation is usually the basis of some kind of motion estimation. Unfortunately, displacement vector estimation using statistical methods is always computationally complex, which might be a restriction in real-time processing. In this paper, we show that displacement vector estimation can be efficiently performed by using cellular neural networks (CNNs). In order to find CNN templates, therefore, we have used the new optimization method of iterative annealing.