Realistic Story Visualization from Online Photo Collections

A complete framework used for creating illustration of a given story is presented in this paper.Guided by a short story and simple sketches, users can create visual story representations from online photocollections. Based on natural language processing method, the framework decomposes a given input storyinto suitable units, and constructs meaningful keywords by which the corresponding images can be automaticallydownloaded from the Internet. An automatic image extraction method based on saliency region isposed for image foreground filtering. Besides, a consistency optimizing method between neighboring scenesis adopted to adjust inconsistent problems. Experimental results indicate that our framework can be used byany user to produce personalized picture series with specified story, even without art background. The workcan be used for children＇s cognition, director＇s creation and some other fields.