Review and Experiment of the Different Methods for Moving Object Extraction from a Live Video

Moving object extraction is a critical stage in a measurable 2D and 3D photogrammetric surveillance systems. In order to keep the frame rate of the live video, the chosen automatic method should be fast enough. Moreover, it is important to extract the moving objects precisely to be able to increase the accuracy of the measurements. Prior to the boundary extraction stage, the moving objects must be detected and separated from each other in each frame of the live video. In this paper, the various spatial domain (e.g. background subtraction) methods including intensitybased and feature-based strategies have been introduced for the purpose of moving object detection. Following the detection step, some utilized methods for object separation have been noted. Furthermore, some useful approaches of error reduction and accurate boundary extraction of detected moving object have also been noted. In addition to the spatial domain methods, some frequency domain methods (e.g. Wavelet decomposition) of moving object detection are also considered in this paper.