Image multi-level classification method and system based on cascaded mean vector comprehensive scoring

Abstract

The present invention relates to the field of image classification, and provides a multi-level image classification method and system based on cascaded mean vector comprehensive scoring, including: constructing a sample image feature vector set through a sample image set, and the sample image feature vector set includes multiple feature categories; Calculate the mean vector of each feature category in the sample picture feature vector set to obtain the mean vector set; extract the picture feature vector of the picture to be classified; iteratively calculate the difference between the feature vector of the picture to be classified and the mean vector of each feature category in the mean vector set The Euclidean distance is obtained to obtain the Euclidean distance set; the classification result of the image to be classified is obtained by calculating the Euclidean distance set. The invention can improve the accuracy of multi-level classification, effectively solve the problem of low classification accuracy caused by not considering the closeness between classes, and the problem of low classification accuracy caused by unbalanced sample distribution; effectively solve the problem of difficult updating of sample warehouses, and greatly improve Reduce the cost of updating the sample warehouse.

Description

technical field [0001] The invention relates to the field of image classification, in particular to a method and system for multi-level classification of pictures based on cascaded mean vector comprehensive scoring. Background technique [0002] In the process of grassroots social governance work, grid personnel need to fill in the event types when they report issues such as inspections and visits, basic data collection, information reporting, conflict investigation and mediation, and event types are often manifested as multi-level cascading classifications, and The three-level classification is the most common. There are many types of incidents in grassroots social governance, and the classification is detailed. Taking Nanshan District of Shenzhen as an example, incidents can be divided into more than 20 first-level classifications, more than 50 second-level classifications, and more than 300 third-level classifications. This makes the grid It is difficult for the personne...

AI Technical Summary

Problems solved by technology

[0007] The main purpose of the present invention is to solve the technical problems in the prior art that the classification results are easily confused due to lack of consideration of the closeness between classes, and the unbalanced distribution of samples will affect the accuracy of the classification model

Images