Video-image segmentation has a broad application in the field of video-image processing and analysis. The accuracy of segmentation directly affects the validity of the continuous tasks, so it has important sense. Low-light environment is a familiar environment for capturing video. The segmentation in this environment is more difficult. There are few studies in this field nowadays.This paper studies the segmentation in low-light environment thoroughly. Firstly, it generalizes the cause and characters of low-light environment systematically, and presents the normal analysis method about the video-image segmentation in this environment. Then, it proposes a fast segmentation algorithm about region-OTSU based on optimal formula and flow to deal with vein video-image; and proposes a fast segmentation algorithm about two-dimension entropies based on fast recursion and search strategy to deal with bill video-image. Validated by many experiments, both of the two algorithms can decrease computational time when it provides good quality segmentation, and can meet the need of fast segmentation in low-light environment very well. |