Research Of Image Annotation By The Joint Modeling Of Object And Motion

With the rapid development of internet application, the images on internet are growing at an exponential rate. Therefore, how to effectively manage and retrieve large scale Internet images put forth a great challenge. These images usually have surrounding text to describe the scenario, so how to select the most suitable words from surrounding text to describe the object or motion expressed in images has become a hot topic recently.Traditional approaches often learn a generative model to denote the occurrence probability between visual objects or motion and their corresponding annotated tags by appearance features, and the learned model is then utilized to recognize persons or actions in a new image outside training samples. However, these approaches just focus on persons or actions, but pay little attention to the relationship between them. In this paper, we propose an approach using SIFT feature of nine position from face representing human and human body probability distribution representing motion to establish an object & motion joint generative modelMoreover, all of existing approaches neglect the grouping effect of high-dimensional features inherent in images. In fact, different kinds of heterogeneous features have different intrinsic discriminative power for image understanding. For instance, the features extracted from arms are most discriminative to human waving motion. The selection of groups of discriminative features for motion recognition is hence crucial. In this paper, we propose an approach to select discriminative subgroup visual features from high-dimensional pose features by Group LASSO during the learning of generative model in order to boost the motion recognition. Experiments show that the proposed approach in this paper can obtain better performance for the recognition of motions with large pose change.