Approach to enhance 3D object retrieval system by considering various characteristics

Recent development and innovations in advanced 3D scanning mechanism has boosted the modelling of 3D objects for several domain related applications. Here we propose a framework for 3D Object retrieval process. Hence for the fruitful retrieval from the warehouse efficient searching techniques are required. In this paper, combination of different methods which are rotation estimation, feature extraction, pose normalization. We initiate the retrieval process by rotation estimation wherein plane reflection symmetry and rectilinearity is used. Following the pose normalization stage, extraction of three types of initial level descriptor is considered collectively than individual descriptor. These factors revamp the retrieval efficiency. Along with this some other factors that affect the retrieval performance are feature selection, preference for appropriate dissimilarity metric and reduction of initial low level descriptors dimensionally. Thus, proposed system facilitates the constitution of various characteristics to enrich 3D object retrieval.