Traditional and modern strategies for optical flow: an investigation

摘要

Optical Flow Estimation is an essential component for many image processing techniques. This field of research in computervision has seen an amazing development in recent years. In particular, the introduction of Convolutional NeuralNetworks for optical flow estimation has shifted the paradigm of research from the classical traditional approach todeep learning side. At present, state of the art techniques for optical flow are based on convolutional neural networksand almost all top performing methods incorporate deep learning architectures in their schemes. This paper presents abrief analysis of optical flow estimation techniques and highlights most recent developments in this field. A comparisonof the majority of pertinent traditional and deep learning methodologies has been undertaken resulting the detailedestablishment of the respective advantages and disadvantages of the traditional and deep learning categories. An insightis provided into the significant factors that affect the success or failure of the two classes of optical flow estimation. Inestablishing the foremost existing and inherent challenges with traditional and deep learning schemes, probable solutionshave been proposed indeed.