Abstracting Videos Using Gaussian Pyramids

Video and digital cameras provide relatively low resolution images, covering a limited field of view. Both the lower resolution and the limited field of view problems can be overcome by combining several images into an extended image mosaic.

The paper investigates the problem of how information contained in multiple, overlapping images of the same scene may be combined to produce images of superior quality. This area, generally titled image mosaic, offers the possibility of reducing noise, extending the field of view, removing blur, increasing spatial resolution and improving dynamic range. As such, this research has many applications in fields as diverse as forensic image restoration, computer generated special effects, video image compression, and digital video editing.

A coarse-to-fine method is used to produce better estimates. Images are put through the following pipeline. They are first smoothed with a 5 by 5 Gaussian filter with a standard deviation of 1.0. For each pair of neighboring images, they are registered under a coarse-to-fine hierarchy using the Gaussian pyramid [34] to produce better estimate. At the coarest level the complete graph is built. The shortest path is found by considering all nodes. This path is interpolated and the optical flow is determined from this path. The new interpolated path is used in next finer pyramid levels.

Throughout this work, the performance of the algorithm is evaluated using real image sequences.