The Design and Development of a Full Range of Display System for 3D Images Based on Android Smartphone

Abstract

With the popularity of smart phones, the demand, that people wish to get a full view of an interested object by smart phones, is increasing. Aimed at the demand, we design and develop such a system, which can all-round displays a 3D object for Android smart phones. This system could be very useful in variety fields, such as tourism, electric mall, educating etc. Introduction Nowadays, the main way for people to get information is through the graphic images. This way, however, with the progress of science and technology, does not meet the need of people any longer. Just like scenery or items in electric mall, people require more details, which mean full range of display of 3D images is the trend of development in the future. Compared to the large side of the computer, smart phones are more convenient to carry. In other words, they are indispensable for people to get information at present. So if a full range of display system for 3D images could be designed, we can enjoy the ‘real scenery’ or observe ‘real items’ in electric mall through the phones in anywhere. Before developing this system, the rapid 3D modeling and full range of display, these two problems, must be solved. On the other hand, this system, based on smart phones, would run lots of energy because of its complicated calculations, which is the fatal flaw for users. So we must use recourses outside of phones, such as cloud computing. Designs and Steps First of all, we use the cloud computing to realize the function of complicated images processing, and it is important to select the cloud. Then, some unnecessary parts in the model would be cut. After that, it will be saved in the server. Finally, 3D models would be showed in phones. Users could stretch or draw, and move the models in any direction, which provides users great convenient. Fig.1 shows the steps below (next page). Building 3D Models Based on Cloud Computing The process of building 3D models is very complicated, that involves large amount of calculation, and runs lots of energy. But using cloud computing could solve this problem, and it is also a way to make the fast generation of 3D model as well. There are many 3D cloud processing websites, but we choose Autodesk 123D Catch (for free) of Autodesk company, which could transform the digital photos to realistic 3D models within the powerful cloud computing. Figure. 1 Steps of the System Design Cutting unnecessary parts and saving in server Showing 3D models on phones Building 3D model using cloud computing International Conference on Education, Management, Commerce and Society (EMCS 2015) © 2015. The authors Published by Atlantis Press 808 Theory and Process of Generation of 3D Models Here is the major theory of the photo-reconstruction in the cloud: Find the characteristics of every photo and match between every two photos. This step is order to distinguish precisely the partly characteristics of items, and match them quickly. But it is difficult actually for many reasons. And the SIFT method referred to David Lowe [1] is the most commonly used in practice. According to the result of matching, the scene information will be calculated using projective theorem. This step is also called Structure from Motion [2] . The accuracy is the standard of the results. It is also called Sparse Reconstruction sometimes. Put the scene information and the origin photos together to get the 3D points cloud of the items in the photos. Multi-view Stereo Reconstruction is available with the scene information. This progress also has a lot of calculation, because of the precision photos. So the execution efficiency is one of the standards to measure the algorism besides the accuracy and integrity of the construction. These points decide the quality of the point cloud. PMVS algorism [3] is the best one for Multi-view Stereo Reconstruction. It is also called Dense Reconstruction. Building the 3D models according to 3D point cloud. It is usually using Poisson Surface Reconstruction to line the point in cloud. Building Models Using Autodesk 123D Catch. Taking photos. It is better to take photos in the same background and in different angles in order to get more details. 40 to 50 photos are enough for a good result, like Fig.2.