An Image Set Classification Method Based on Aggregate Hash Learning

A classification method and image set technology, applied in the field of image set classification based on aggregate hash learning, can solve the problem of not considering the image set set information well, affecting the classification accuracy of image sets, and it is difficult to obtain high-quality instance hashes Code and other issues

CN114596456BActive Publication Date: 2022-07-22SICHUAN UNIV
2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Publication Date
2022-07-22

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses an image set classification method based on aggregated hash learning, comprising the following steps: S1, establishing a hash aggregation model according to a training image set; S2, training the hash aggregation model to obtain a linear hash after the training is completed. Hash function and the instance hash code matrix of the training image set; S3, calculate the hash code of the instance of the image set to be classified according to the linear hash function completed by the training; S4, according to the hash code of the instance of the image set to be classified and The distance between the instance hash code matrices of the training image set is used to obtain the category of the image set to be classified; the present invention solves the problem that the existing hash learning method does not take into account the set information of the image set well, and is difficult to perform well in the Hamming space Keeping the similarity of instances, it is difficult to obtain high-quality instance hash codes, which in turn affects the classification accuracy of image sets.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention relates to the technical field of image classification, in particular to an image set classification method based on aggregated hash learning. Background technique

[0002] Over the years, with the rapid development of multimedia technology, image set classification has been widely used in a large number of real-world applications, such as object classification, video-based face recognition. Compared with traditional single-shot image-based classification tasks, image set classification pays more attention to the generality of the set and the complementarity of samples, which can provide richer information to describe the corresponding topics. The interest in image set classification has been steadily growing in the research community as image set classification can provide more promising performance to overcome the appearance variations of images (e.g. appearance, occlusion, lighting, expression, pose). However, with the increase of dat...

Examples

Embodiment Construction

[0054] The specific embodiments of the present invention are described below to facilitate those skilled in the art to understand the present invention, but it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes Such changes are obvious within the spirit and scope of the present invention as defined and determined by the appended claims, and all inventions and creations utilizing the inventive concept are within the scope of protection.

[0055] like figure 1 As shown, an image set classification method based on aggregated hash learning includes the following steps:

[0056] S1. According to the training image set, establish a hash aggregation model;

[0057] The training image set consists of one or more training subsets of images from each class used for training, and each image training subset consists of different images belonging to the same class.

[0058] Step S1 i...