View Selection of 3D Objects Based on Saliency Segmentation

Getting good viewpoints has been considered important for promoting the efficiency when investigating a model. Many view selection methods therefore have been proposed. In particular, measuring semantic meaning of the model features through segmentation is regarded more effective to get optimal viewpoints. Unfortunately, the semantic meanings of the model are usually obtained through experiences, which are not easy to apply into practices. Given these considerations, a new method is proposed for view selection via saliency segmentation in this paper. Due to the fact that mesh saliency computation can recognize the features of a model in a way that similar to human perception and also a novel viewpoint ranking equation has been developed, our method is more reasonable than existing methods. The results of experiments show the effectiveness of our method, and it is consistent with the results of a user study we conducted.