Remote Sensing Image Retrieval Algorithm Research Based On Tolerance Granular Theory

Due to the "Digital Earth" issue, in the field of information technology, the rapid development of remote sensing technology has made remarkable achievements. This is followed by the increasing use of various satellite sensors and exponential growth of remote sensing data from the network. Since these remote sensing data have many formats, different forms,resolution and functional differences,it requires a very high demand with respect to its management and storage. How to effectively store and retrieve remote sensing data have become a requirement which is need to be urgently solved and a challenging issue. Remote sensing image retrieval technology is the core of image retrieval technology, which in recent years is solving the massive remote sensing image data by key technology of information science, and more and more researchers are concerning this issue. However, according to the practical application, it is found that the content-based remote sensing image retrieval technology is similar to the generic content-based image retrieval technology. However, in the field of remote sensing image retrieval, generic content-based image retrieval technique cannot be directly applied, and requires some special way or method to solve the differences between the remote sensing images and generic images. Therefore, there are still many problems, which need further exploration of scholars.In order to solve the above-mentioned problems, this paper combined the content-based image retrieval technique with the new intelligent algorithm in the field of artificial intelligence. The paper investigated the tolerance granular theory, and deeply studies and analyzes the two theories of the tolerance granular theory. The tolerance granular model and the tolerance granularity space model are applied to the field of remote sensing image retrieval. Main conclusions are as follows:(1) The region matching algorithm is improved. In this paper, the significant factor of James Wang's comprehensive region matching algorithm is studied. The weight ratio of the area ratio, the image position and the image brightness is improved from the simple area ratio.The improved algorithm is improved compared with the original algorithm in the field of remote sensing image retrieval.(2) The remote sensing image retrieval algorithm based on the tolerance granular model is proposed according to triad. Compared with the improved integrated region matching algorithm, The average precision of this algorithm is 0.7758, which is 6.66% higher than that of the improved IRM, which satisfies the user's requirement.(3) A remote sensing image retrieval algorithm based on tolerance granular space model is proposed. According to the four-element model of tolerance granular space,a three-layer remote sensing image tolerance granular space model is established and the similarity calculation is carried out. The quasi rate is 0.70125, and the average precision is 13.875%higher than that of the region matching algorithm. It can be seen that the algorithm has improved the precision in remote sensing image retrieval.In this paper, the application of tolerance granular theory to remote sensing image retrieval and achieved the desired effect, the experiment proved the effectiveness of the proposed algorithm.