Research On 3D Model Retrieval

With the rapid development of multimedia technology,3D models have been widely used in diverse fields,such as virtual reality,3D modeling and 3D games.The number of 3D models grows with the exponential method.It is a problem to be salved how to effectively and efficiently retrieve 3D model from the Vast amounts of model.Though traditional context-based methods is easy to implement and speedy in time,manually annotated procedure can be easy affected by subjective ideas and takes too much labor cost,which severely limit their practical applications.While content-based methods works superior in both performance and cost,becoming a burgeoning emerging technology.we will explore view-based methods in 3D model retrieval.In this paper,we will focuses on the view-based 3D model retrieval,then raised on two new method about 3D model retrieval.1)Graph-based 3D model retrieval:using the virtual camera to obtain a set of three-dimensional model of two-dimensional view from a different perspective,Zernike moments feature is representative of the view;Graph cut method is used to extract the representative views;Finally,the graph matching method is used to calculate the similarity between3 D model.2)Topic-model-based 3D model retrieval: Firstly,we extract the SIFT features from the views to construct the Bow features;then use the LDA topic model to infer the topic distribution.Finally,by calculating the distant of the topic distribution between models measure the similarity.In this paper,we selects the more popular 3D database to evaluating our method,and compare with some classical algorithm.The final experimental results demonstrate the effectiveness of our method.