Automatic commodity image classifying algorithm based on local feature multistage clustering and image-class distance computation

Abstract

The invention relates to an automatic commodity image classifying method and provides an automatic commodity image classifying method based on vision information. The automatic commodity image classifying method comprises the following steps of: extracting local features of an image, and realizing class recognition by adopting a hierarchical clustering and image-class distance computing method. The automatic commodity image classifying method has the advantages that the automatic commodity image classification can be realized according to vision contents of the commodity image; by adopting a method of describing a subcluster by using local features of each class of images, an exhaustion searching problem in the image-class distance computation is greatly reduced and the computing efficiency is greatly improved; and when the distance between image blocks is computed, position information of the features is taken into full account, so that the classifying effect can be improved.

Description

technical field [0001] The invention relates to an automatic classification method for commodity images, in particular to a commodity image classification algorithm based on local feature hierarchical clustering and image-class distance calculation. Background technique [0002] With the development of the Internet, e-commerce is becoming more and more popular. E-commerce websites need to mark the products sold online to facilitate users' search. But "a picture is worth a thousand words", the traditional method based on manual annotation is not only time-consuming and laborious, but also difficult to be accurate and complete. If you set a picture classification filter in the website, it will undoubtedly be convenient for users to browse. Image automatic classification technology based on visual information can facilitate merchants and users, such as automatic product labeling and assisted image retrieval. [0003] Automatic image classification based on visual information...

AI Technical Summary

Problems solved by technology

However, since this method is based on exhaustive search, when the number of labeled images is large and the number of categories is relatively large, the running time will become larger and larger, and it is difficult to meet the requirements of practical applications.

Images