An Integrated Color and Texture Feature Extraction Algorithm

Guangwen Zhang,Lei Yang,Fan Zhang

DOI: https://doi.org/10.1109/iccsnt.2012.6526038

2012-01-01

Abstract:With the rapid development of Internet, the category and quantity of the pictures increase sharply. Some are better to use color retrieval method, but some are more suitable to texture retrieval. A single retrieval method sometimes can not fully describe all the characteristics of the image. A single feature extraction method has difference with the human eye visual, and two totally irrelevant image feature vector may be located very close to, so that it will cause errors in the retrieval process. However, the probability of two different images obtaining similarity measure in the other characteristics is also very small, so a combination of multi-feature query is produced. On this basis, a multi-feature combination query allows customers to express their query requirements flexibly. The basic process is: firstly, combination of the query is decomposed into a number of queries based on individual characteristics, and all the results of image retrieval is returned by a single characteristic, then the system merges the similarity measure of each feature in accordance with a specific algorithm. Reorder the integrated similarity measure in the press after the merger of these images, and choose the n images with maximum similarity as the search results returned to the user. The experimental results show that this algorithm is better for multi-feature image retrieval results.