Đang chuẩn bị liên kết để tải về tài liệu:
Abnormal posture detection in an omnidirectional camera based surveillance system

Không đóng trình duyệt đến khi xuất hiện nút TẢI XUỐNG Tải xuống

The background subtraction method is first applied to extract targets. Then principal component analysis (PCA) and the change of height of a personal trapezoidal bounding box are used to detect fall down and crouch postures. Several experiments on various environments based on the proposed approach were conducted and evaluated. Stable detection results were obtained to show the feasibility of the proposed system. | Journal of Automation and Control Engineering Vol. 4, No. 6, December 2016 Abnormal Posture Detection in an Omnidirectional-camera-based Surveillance System Din-Chang Tseng and Chien-Hung Chen The Institute of Computer Science and Information Engineering, National Central University, Jhongli, Taiwan Email: tsengdc@ip.csie.ncu.edu.tw, chienhung66@gmail.com it is quite sensitive to changes in ambient light, and cannot overcome the problem of camera movement. Optical flow method is using a variation on the estimation of image pixels to infer the position of moving objects. This method can detect when you move the camera still moving object independently, but the disadvantage is fairly complex calculation, if there is no specific hardware, cannot be applied in real-time systems. Background subtraction is quite widely used method of dividing the movement area in the image sequence. The simplest and most common is the use of a background subtraction few seconds of time to build a normal distribution in the background of each pixel, then sent to new imaging sequences and background model for comparison, and set the threshold to distinguish between the foreground and background points many systems take advantage of such a method to detect the movement of the pixel [7]-[14]. Collins et al. [15] combined the temporal difference and background subtraction method to detect a moving object. Using the time difference method to detect the moving area, then use in the mobile area background subtraction to obtain complete mobile pixels. Horprasert et al. [16] proposed a method to detect moving objects from a static color background images, background model to establish a reference image using a statistical approach. Javed et al. [17] proposed a gradient-based background subtraction method, the feature vector from the gradient size and the direction of the composition gradient, based on the establishment of a gradient of the background model. Lu and Tan [18] uses HSV color space .