Image semantic retrieving method based on visual attention model

Abstract

The invention provides an image semantic retrieving method based on a visual attention mechanism model, which is driven by data completely, thus understanding the semantics of images from an angle of a user as far as possible under the condition of no need to increase the interactive burden of the user, and being close to the perception of the user so as to improve the retrieving performance. The image semantic retrieving method has the advantages that: (1) a visual attention mechanism theory in a visual cognition theory is introduced into image retrieve; (2) the method is a completely bottom-up retrieve mode, thus having no need of user burden brought by user feedback; and (3) obvious edge information and obvious regional information in images are simultaneously considered, the mode of retrieval integration is realized, and the performance of image retrieval is improved.

Description

technical field [0001] The invention relates to image recognition and retrieval technology, in particular to an image semantic retrieval method. Background technique [0002] With the rapid development of multimedia technology and Internet technology, digital images have become a widely used media. In recent years, the rapid popularization of digital cameras and mobile devices that can take pictures has made it easier to obtain digital images. The number of images that people come into contact with and need to process every day has shown geometric growth, and the scope of application has also greatly expanded. Facing such a large-scale image resource, how to effectively organize and quickly retrieve them has become an urgent problem to be solved. Because the image is different from the text, the text itself can explain the content of the speech, but the image needs the help of human subjective understanding to explain its meaning, so the retrieval of the image is much more ...

AI Technical Summary

Problems solved by technology

The "many-to-many" matching strategy is used in the matching, that is, the integrated region matching (Integrated Region Matching, IRM) algorithm is used to measure the similarity of the image. It should be pointed out that since all image features and matching are used. The segmented area, but no selection of the area. Since most areas are not of interest to the user, the whole-area matching leads to a decrease in retrieval accuracy, which still belongs to the global retrieval mode.

[0008] Although region-based image retrieval is closer to the user's query ideas than global-based image retrieval, it also has some problems: First, because image segmentation is still a very difficult topic in the field of computer vision, the existing image segmentation technology It is impossible to guarantee the accurate extraction of objects in the image, so that the segmented area corresponds well to the semantic object; The regions of interest represent the user's query intent, while most of the remaining non-interest regions are irrelevant to the user's query intent

Therefore, the retrieval strategy based on whole-region matching not only cannot reflect the user's retrieval purpose, but also these irrelevant regions are often difficult to match correctly, resulting in a decrease in retrieval accuracy

However, manually selecting the area of ​​interest by the user will virtually increase the workload of the user, and the user is not used to this query method

Images