Product quantization method based on semi-supervised learning

Abstract

The invention discloses a product quantification method based on semi-supervised learning, which is an improved algorithm based on a common Cartesian K-means algorithm, namely a semi-supervised Cartesian K-means algorithm. In the algorithm, a traditional minimum square loss function in a quantization step needs to be replaced with an optimal reverse prediction loss function. According to the traditional semi-supervised learning, the marked data can be directly used for model training, and different from a traditional semi-supervised learning model, the marked data can be used for model training only through Laplacian regularization.

Description

technical field [0001] The invention relates to the technical field of data processing and pattern recognition, and is mainly used for image classification, especially an image processing method for distinguishing different types of objects according to different characteristics reflected in image information. Background technique [0002] In order to overcome the disadvantages of image retrieval algorithms characterized by text keywords, researchers proposed the concept of Content-Based Image Retrieval (CBIR). It mainly uses the shape, color, texture and other characteristics of the image to match the relevant information of the image in the database to obtain a similar picture, thus avoiding the tedious and complicated work of artificially classifying and calibrating the image content. [0003] Existing large-scale search engine platforms have launched image retrieval services characterized by image content, such as Google’s “Search by Image” and Baidu’s “Baidu Recognition...

AI Technical Summary

Problems solved by technology

[0009] The above algorithms have been significantly improved in the study of product quantification algorithms, but all these algorithms still belong to the unsupervised learning framework, which may clearly limit the performance of these algorithms

Images