Recently, video surveillance system using high resolution cameras is increasing. For such video surveillance system, there are two important requirements: high-reliability in detection and network-friendliness. However, in case of surveillance in night, image enhancement should be performed but enhanced video quality will be degraded. Especially, encoded video will boost quality degradation such as block noises. Therefore, when rate control for enhanced videos is processed, quality degradation influenced by image enhancement should be considered. In this paper, to research the relation between image enhancement and encoding, we evaluate the difference of videos that are enhanced or not, and we found that image enhancement boosted the quality degradation. Also, we carried out human detection in enhanced videos and found that image enhancement lead to high accuracy of human detection when image contrast is low. Moreover, based on these results, we show an example of control parameters of encoding rate and image enhancement strength.
View full abstract