Article Preview
TopImage File System Resembling Memory Storage
With the development of WEB2.0 and popularity of social network and online shopping, the number of image files is increasing exponentially on the Internet. The large number of images makes the storage challenging, which requires not only the efficiency but also the robustness. There have been many distributed file systems proposed and implemented over the years. However few of them meet previous demand.
Traced back in 2000, a framework of higher order databases (Schek, & Weber, 2000) was proposed and it was applied to multimedia information systems, with which a networked image retrieval and management system was built. In recent years, there are also some research works about distributed image file system. Three basic characteristics of distributed file system were analyzed (Yan, SHUAI, & Ping, 2011), including the classification of multi-node servers, the data distribution, and the communication between the multi-node servers. After the analysis, a convenient and flexible framework was proposed for distributed storage system on a cluster of commodity machines. Beside this framework, an unstructured database management system named Advanced Unstructured Data Repository (AUDR) was introduced (Liu, Lang, Yu, Luo, & Huang, 2011) which is designed to manage massive and various types of unstructured data including text, image, audio and video. There are also other image storage systems, for example a Distributed Image Retrieval System (DIRS) is established (Zhang, Liu, Luo, & Lang, 2010) in which images are retrieved in a content-based way, and the retrieval among massive image data storage is speeded up by utilizing MapReduce distributed computing model. A framework called OBSI was also presented (Yu, Chen, & Bei, 2007) which stands for an object based storage system for storing and retrieving images of online album services. OBSI provided high efficiency and throughput in image storage and retrieval services.
Some distributed systems for the specified type of images were also proposed. A secure and cost effective distributed file system, JigDFS, was presented for archiving medical images/data (Bian, Seker, & Topaloglu, 2010). An integrated data management system was developed in order to store and manage a large amount of ground and indoor image data with high resolution (Yuan, Xun, Wang, Yuan, Dan, & Diao, 2011). Cloud computing technology was used to provide storage and processing capabilities for extremely large image-based datasets and photographic editing workflows, and it demonstrated the performance gains over traditional system (Malensek, & Pallickara, 2012).
There were also some research works focusing on the image retrieval and image sharing from distributed file systems. Ménard examined image retrieval within two different contexts: a monolingual context where the language of the query is the same as the indexing language and a multilingual context where the language of the query is different from the indexing language (Ménard, 2010). A service-oriented remote sensing image data sharing prototype system is realized (Wang, Wang, & Wang, 2011), which is technologically available for the integration of heterogeneous and geographically distributed image storage and management system.
Previous mentioned distributed file system couldn’t provide high efficient and robust image storage and retrieval. According to the state of art research, humans can see and name thousands of distinct object and action categories (Torres, Jaime, & Ramos, 2013), so it is unlikely that each category is represented in a distinct brain area (Huth, Nishimoto, Vu, & Gallant, 2012). A more efficient scheme would be to represent categories as locations in a continuous semantic space (Xu, Zhi, Liang, Mei, & Luo, 2014; Xu, Yan, Liu, & Mei, 2013). Thus we proposed a distributed image file storage system based on cognitive theory (Wang, 2014; Li, Peng, Chen, Gui, & Song, 2014).