Research Of Positioning System Based On Image Processing

In recent years, research on image processing technology gains more and more researchers’ attention, Both in industry and academia, has a number of potential application scenarios and the significance of the research of image processing technology. But the traditional image processing techniques are limited to a specific complex algorithm which gets important information of the image. With the rise of big data technology, the combination with image processing technology and big data technology is the trend. Big data technology enrichs image processing technology and achieves good results. The positioning system is also one of the research hotspost, but the recent postioning technology is based on the special navigation system such as GPS which is difficult to deal with the situation lacking of navigation system. And the degree of precision of GPS navigation system also needs to be further improved. In 2011, Berkeley university put forward the "Location-Based Image Retrieval For Urban Environments", in order to achieve the postioning system through image processing technology. This paper is through the study on city environment positioning system, combining with the image processing technology and big data, put forward a more efficient solution compared to Berkeley university. According to this problem, we put forward the following solutions, the research results are as follows:I. A detailed analysis of the related theory and realization method of SIFT algorithm; For the features and difficulties of urban environment, we analyse the shortcomings of simple use of SIFT features.II. Through the combination with bag-of-words and SVMs which is based SIFT features, we achieve the image classification, Using manual annotation to achieve the corresponding image and map, and the implementation of positioning system.III. We analyse the advantages and disadvantages of matching system based on SIFT features and classfication system based on bag-of-words in detail. Through the experiment, we prove that the classification system based on bag-of-words has a good performance both in accuracy rate and efficiency.To sum up, this paper systematically studies the related research method of image processing technology, according to the features of the urban environment, we put forward the bag of words model classification system based on SIFT features, and analyse the advantages of it compared to matching system based on SIFT feartures. According to the algorithm in this paper, we has carried on the corresponding simulation experiments, and prove the effective and accuracy of the algorihthm in this paper.