A real-time image clustering method

Abstract

The invention discloses a real-time image clustering method, comprising the following steps: A. batch input images, and obtain image global descriptors through edge feature extraction; B. obtain image local feature descriptors through low-resolution image feature extraction; C. Input the image global descriptor and the image local feature descriptor; D. Create a three-layer self-organizing mapping neural network, and use the image global descriptor to select the first N competing layer neurons to enter the activation state; E. Calculate the image local feature descriptor and each The distance between each nerve ending under each activation neuron is obtained, and the activated neuron that responds successfully as a whole is obtained; F. According to the number of activated neurons that respond successfully as a whole, the clustering of images and the learning or combination of activated neurons are performed, or Create new competition layer neurons using image global descriptors and image local feature descriptors. The present invention improves the accuracy of the image clustering result and the real-time performance and stability of the clustering process as a whole.

Description

technical field [0001] The invention relates to the technical field of photogrammetry and remote sensing mapping, in particular to a real-time image clustering method. Background technique [0002] In recent years, the wide application of information network technology has continuously promoted changes in lifestyles. The Internet, the Internet of Things, knowledge services, and intelligent services have become an indispensable part of people's lives, thus forming a huge network of micro-sensors, resulting in Difficult to quantify unstructured image data. These image data are complex in type, huge in volume, strong in timeliness, and have obvious big data appearance, and have become an important research object. The first step to deal with these imprecise and unstructured image big data is to carry out autonomous clustering among images to find image sets with similar content in the same target area. The clustering between images can be roughly divided into two steps: one i...

AI Technical Summary

Problems solved by technology

In addition, the calculation accuracy and real-time performance of the current clustering algorithm still need to be improved. How to better introduce the neural network into the current algorithm to improve the clustering accuracy and real-time performance is a problem worth studying.

Images