In this paper, we present a new strategy, a joint deep learning architecture, for two classic tasks in computer graphics: water surface reconstruction and water image synthesis. Modeling water surfaces from single images can be regarded as the inverse of image rendering, which converts surface geometries into photorealistic images. On the basis of this fact, we therefore consider these two problems as a cycle image‐to‐image translation and propose to tackle them together using a pair of neural networks, with the three‐dimensional surface geometries being represented as two‐dimensional surface normal maps. Furthermore, we also estimate the imaging parameters from the existing water images with a subnetwork to reuse the lighting conditions when synthesizing new images. Experiments demonstrate that our method achieves an accurate reconstruction of surfaces from monocular images efficiently and produces visually plausible new images under variable lighting conditions.

Abstract. Review similarity computing is used to judge whether the content of online reviews is related to the products. It is an important prerequisite to judge the usefulness of reviews, and it is also an important basis for the classification and sorting of product reviews. This paper combines the VSM, TF-IDF algorithm and cosine similarity algorithm to build the model of similarity computing between the product online reviews and product features, and to build the process framework of review similarity computing for enterprises. Besides, this paper also verifies the model's effectiveness and correctness based on real online review data of E-business. The experiment results show that the process model can be used to quantify the similarity between reviews and product features, and the similarity results also have a good effect on the application of the review sorting.

No abstract

No abstract