Scheme for Evaluation and Reduction of Motion Artifacts in Mobile Vision Systems

Abstract

Artifacts like motion blur are a common problem for vision systems on mobile robots, especially when operating under low light conditions or when using high-resolution sensors. In this contribution we present a scheme for estimating the degree of motion artifacts, especially motion blur, present in a stream of individual camera images. A single quality estimate is derived per frame using data from an inertial measurement unit. Considering limited image processing capacity of resource-constrained mobile robots, we show a number of data processing strategies that are based upon the idea of congestion control by adaptive image rejection.