The Research And Implementation Of Content-based Image Retrieval Technology For The Image Of Cultural Relics

China has rich cultural resources, but the preservation of these cultural relics is very difficult. In order to have a better appreciation and protection of these resources, the establishment of a complete digital heritage conservation system is imperative.As an important part of cultural relics digital process, image retrieval technique has become a new research hotspot. Heritage image itself has the characteristics of large amount of data, high dimensions and resolution. So traditional retrieval methods cannot achieve a fast and accurate retrieval of massive cultural image. Then, Content-based image retrieval(CBRI) came into being, and quickly became the technology upstart in the field of image processing and computer vision.In this paper, the key technologies of CBIR were explored and research. We also built an image retrieval model with good performance. The main work includes:(一) We studied the extraction algorithm of image low-level visual features. Through the comparison and analysis of a large number of experimental data, we developed a comprehensive use of color features and LBP-HF texture features, as a multi-feature fusion to achieve a better image retrieval result. Experiments show that the retrieval accuracy is higher than that of the single feature search.(二) The similarity measure of images has been explored. Since the importance of image color information and texture information is different, when using different characteristics at the same time, we need to set appropriate weight for them. In this paper we proposed a new method, we use the sum of different weighted feature similarity as the integrated similarity metrics.(三) we combined image retrieval technology which based on image low-level visual features with the relevant feedback technology to avoid the "semantic gap" caused by the lack of high-level semantics. Through the continuous learning of users’ feedback information, adjust the search strategy, and then improve the performance of cultural image retrieval system.