An Efficient Video Indexing and Retrieval Algorithm using Ensemble Classifier
N. Gayathri, K. Mahesh · 2019
In the recent times, video are swamping over Internet in applications such as social networks, business multimedia systems and websites. These applications adopt video as foremost significant form of information or communication. Videos can be generally accessed as a whole and it will not be indexed in visual content. For instance, videos are generally uploaded as manually, short clips, cut clips with user provided tags, keywords and annotations for retrieval. In this investigation, a classification model has been designed to categorize the video based on similarity and dissimilarity. This classification model address two limitations: it eliminates the intervention of human in video retrieval. Next is time complexity of retrieving video content is lesser in contrast to prevailing state of the art approaches. In this work, initially, a series of images from any sample dataset or from available video is considered. Subsequently, perform feature extraction using two approaches termed as Independent feature extraction and principal feature selection based dimensionality reduction. Finally, classification for video retrieval is performed with logitboost ensemble classifier. Simulation is carried out in MATLAB environment, performance metrics like accuracy, precision, recall and time complexity is attained. Here, the proposed model shows higher accuracy, precision and recall value, whereas time complexity is reduced in contrast to prevailing approaches. The proposed model shows better trade of than existing approaches.