Detection of moving objects using a Gaussian mixture model

Computer has been used for so long to watch crowded public domains such as vision community, shopping malls requiring sensitive security and precaution, banks, motorways, schools and hospitals. Increase in the data processing rate is become a considerably popular subject that the use of computers in subtracting contains from video files. Although the use of computers to keep digital video files is considerably common, its use for subtracting the necessary information from the digital videos is brand new. The most important rudiments obtained from digital video files are mobile objects. This study is an authentic work about background modeling and mobile object extraction-algorithm in digital video files. In the study, not only a pixel worth is independently handled from its neighbors in time and space, but also it is taken into account that backlighting of an image can be variable. For these reasons, for every pixel in a fixed sized image is applied ?Gaussian Mixture Distribution Model?. Generally, a Gaussian Mixture Model is described with verifying data that is the total of a few Gaussian distribution which have both differential average and variance values and different weight within the model. Statistical data analysis is done with EM Algorithm. Experiments about various digital videos are done. Test results of these experiments show that recommended method productively determines the detection of mobile objects. Key Words: Background Extraction, The Detection of Mobile Objects, Gaussian Mixture Model