Reconstruct Clear Image for High-Speed Motion Scene with Retina-Inspired Spike Camera - Supplementary Material

Supplementary material for the paper: 'Reconstruct Clear Image for High-Speed Motion Scene with Retina-Inspired Spike Camera'

PAPER ABSTRACT: Conventional digital cameras typically accumulate all the photoelectric information within an exposure window to form a snapshot image. It requires the scene to be quite still during the exposure interval, otherwise it would result in blurry image for the moving objects. Recently, a retina-inspired spike camera has been proposed and shown great potential for recording high-speed motion scene. Instead of capturing the visual scene by a single snapshot, the spike camera records the dynamic changing process of light intensity continuously. Each pixel on spike camera sensor accumulates the incoming photons independently and persistently, and fires spikes once the dispatch threshold is reached, producing a continuous stream of spikes recorded at very high temporal resolution. To recover the dynamic scene from the captured spike data, we present an image reconstruction approach for spike camera. In order to generate high-quality reconstruction, we investigate the temporal correlation along motion trajectories and exploit it via an adaptive temporal auto-regressive model. In particular, we present a hierarchical motion-aligned temporal filtering scheme, combining short-term filtering with long-term filtering to take advantage of long-term temporal correlation with much lower model complexity. Experimental results demonstrate that the proposed method produces high-quality images for high-speed motion scenes, improving the reconstruction images in terms of both objective and subjective qualities.

This dataset provides a set of results and demos for our work on spike camera image reconstruction.