e., Calyptogena soyoae) at 1,100 m depth off Hatsushima Island, in Sagami Bay (34��59.97��N, 139��13.69��E; central Japan). The video-camera was continuously new product acquiring images in a time-lapse mode (i.e., a frame each 4 s) under a constant source of illumination (i.e., six white-light lamps). Videos were stored on VH-S videotapes and their processing in relation to behavioural rhythms was never attempted before. We selected long lasting continuous 10 years-old footage for the technological challenge of performing automated video-image analysis on videos lacking digital standards. The footage was digitized and partitioned into frames at a rate equivalent to the frequency of video recording. Footage processing and video-image analysis were both carried out with MatLab 184.108.40.206.
Video-image AnalysisMotion detection procedure Inhibitors,Modulators,Libraries identified animals based on their displacement through consecutive frames [14-17]. In this process, the quality of extracted information depends upon several contingent factors typical of the deep-sea context . Firstly, the detection of movement depends upon the rate of image acquisition in comparison to the speed of animals’ motion. Secondly, a source of continuous white illumination, gradually decreasing over the distance (i.e., within 3�C4 m) is always present during filming operations. This may impair animal detection depending on its positioning within the camera field. Thirdly, consistent water turbidity is often present (i.e., high-contrast organic debris as ��marine snow��), creating difficulties in the automated identification of moving animals.
Inhibitors,Modulators,Libraries Fourthly, different species possess different shapes that are also variables according to animal displacement within the camera filed.In spite of all these considerations, an automated video-image analysis protocol was developed according to two major steps: (1) animals’ motion detection, by means of image extraction; (2) animals’ recognition within different species categories, by multivariate morphometric techniques such as Fourier Descriptors (FD) and the Supervised Standard Inhibitors,Modulators,Libraries K-Nearest Neighbours (KNN) analyses. A flow chart specifying Inhibitors,Modulators,Libraries the different steps involved in the automated procedure is illustrated in Figure 1.Figure 1.Flow chart representing the different steps involved in the automated procedure.2.3.
Animals’ Motion DetectionAutomated video-image analysis for the tracking of movement was based on combined frame subtraction and a multiple filtering procedure (Figure 2).Figure 2.The consecutive stages of automated video-image analysis for the tracking of movement. Entinostat A frame is firstly subtracted by its consecutive (A). The image without sensible objects is used for background subtraction (B), which occurs within a region of interest …We applied a high throughput screening simple algorithm in order to subtract the current image for a background image of reference.