Semantic Tolerance-Based Image Representation for Large Image/Video Retrieval

The nature of the concepts regarding multimedia in many domains is imprecise, and the interpretation of finding similar media is also ambiguous and subjective on the level of human perception. To solve these problems, in this paper, semantic categories of images or key frames which are extracted for representing the segments of a video, and the tolerance degree between the categories are defined systematically, and the approach of modeling tolerance relations between the semantic classes is proposed. Furthermore, for removing the induced false tolerance in the produce of using semantic tolerance relation model, the method of un-tolerating is introduced in image/key frame representation. On the other hand, a diagram of semantic tolerance-based image/video automatic representation is described, and the structure of large image/video retrieval using image/video semantic representation is proposed. We apply the proposed approach to the representations of images regarding the nature vs. man-made domain, human vs. non-human domain, and temporal domain, and show the categorization results of using and not using semantic tolerance relation model. Furthermore, the mechanism of the semantic representation and retrieval for large image/video data proposed in this paper is compared with the state-of-the-art methods. The results show the effectiveness of proposed method.