Real Time Semantic Adaptation of Sports Video with User-centred Performance Analysis

摘要: Semantic video adaptation improves traditional by taking into account the degree of relevance different portions content. It employs solutions to detect significant parts and applies compression ratios elements that have importance. Performance semantic heavily depends on quality precision automatic annotation, whether it operates in strict or nonstrict real time, codec which is used perform at event object level. should consider effects errors extraction objects events over operation subsystem, relate these preferences for program, been decided user. In this paper, we present time annotation sports introduce two new performance measures: Viewing Quality Loss Bit-rate Cost Increase, are obtained from classical PSNR Bit Ratio, but results with user’s