A Survey on Various Techniques of Semantic Object Extraction in Videos

Video-based applications are largely being used these days. These applications include video surveillance, criminal detection and sports video analysis etc. In particular, Object based representation consists of decomposing the video content into a collection of meaningful objects. There are many literatures present on developing technique for efficiently querying videos on their content. The content of the video can be basically defined as the objects and the iteration of the objects. So the Object identification and classification has become challenge these days. Object Extraction strategies can be broadly categorized in three parts Manual, Semi-automatic and Automatic. In order to make object extraction technique Automatic, Algorithms are applied on extracted features of images for classification. Genetic Algorithm is used to automatically classify the objects. To find the optimal solution from genetic algorithm, it is required to maintain the larger population size which is not cost effective. There is another approach to classify the candidate object which is Fuzzy C-means based GA. In this paper, object extractions strategies are reviewed followed by discussion on Fully Automatic approaches of Object classification.