This paper presents a novel system for a real-time no-reference video quality monitoring on a commercially available multi microprocessor platform1. System design stage incorporates, firstly, automated statistically based selection of appropriate objective measures from a large predefined set for characteristic artifacts. Secondly, this reduced set of objective measures is used in conjunction with subjective assessments for derivation of a non-linear model which can then be used for future predictions. Different sets of relevant objective measures are used for video material with various contents, and hence for derivation of content based non-linear video assessment models. The system is capable of capturing live video, selecting appropriat...