Image classification method

Abstract

The invention provides an image classification method, and particularly relates to an application method of concept cognition in the field of image classification. The image classification method based on the concept cognition process is divided into two modules: a cognition training module used for learning an image concept and finally generating a concept tree for realizing an image classification task; and the image classification module that is used for carrying out image classification by using the concept tree obtained by the training module. The method comprises the following steps: describing an image feature FD, generating an image form background K, generating a necessary attribute set Base by an image set form background K, constructing a coarse level of a concept tree CT by necessary attributes, constructing a fine level of the concept tree CT by the image set form background K, updating the concept tree CT, obtaining an optimal concept tree, and classifying by using the optimal concept tree. The problem that formal concept cognition is difficult to apply to the image aspect is solved through an image feature extraction technology, a feature data processing technology and a concept learning method.

Description

technical field [0001] The invention belongs to the technical fields of pattern recognition and machine learning, and in particular relates to an image classification method. Background technique [0002] Since formal concept analysis was put forward, it has become an efficient research tool in the field of data analysis by virtue of its unique relational representation. Formal concept analysis has been successfully applied to recommender systems, data mining and other related research fields. Conceptual cognition is a research direction that has emerged in recent years. It has interdisciplinary characteristics and involves many research fields such as psychology, brain information science, and computer science. Existing research proves that conceptual cognition learning is an efficient information processing method. However, most of the existing research stays at the level of theoretical demonstration. At present, the application of formal concept analysis in the image fie...

AI Technical Summary

Problems solved by technology

Conceptual cognition is a research direction that has emerged in recent years. It has interdisciplinary characteristics and involves many research fields such as psychology, brain information science, and computer science. Existing research proves that conceptual cognition learning is an efficient information processing method. methods, but most of the existing research stays at the level of theoretical demonstration

At present, the application of formal concept analysis in the image field is seldom, the main reason is that the inherent time complexity of the concept lattice cannot effectively deal with the multi-dimensional information in the image.

Images