Image feature extraction method, device, terminal equipment and system

Abstract

The present application provides an image feature extraction method, which first receives an image input by a user; then performs registration on the image input by the user to obtain a registered image; and then constructs multiple structured structures for the registered image sub-image; then adopt the feature model obtained by multi-model training to extract the visual features of each of the structured sub-images; then carry out structural fusion of the visual features of the multiple structured sub-images to obtain structured feature data; finally A model obtained through structured model training is used to perform operations on the structured feature data to obtain image feature data. Compared with the prior art, the image feature data obtained in this application are feature vectors. Since the feature vector and the model maintain the structural characteristics of the image during the training process, the image feature data has higher accuracy and reliability. Recognition, the application in image recognition, especially face recognition, will have higher accuracy, so as to obtain better recognition effect.

Description

technical field [0001] The present application relates to the field of electronic technology, and specifically relates to an image feature extraction method, an image feature extraction device, an image feature extraction terminal device, and an image feature extraction system. Background technique [0002] Face recognition research began in the 1990s. At the beginning, there were eigen face method to describe face with image principal components and fisher face method to describe face image with distinguishing features; after entering this century, based on LBP Gabor's face local feature description method and boosting-based discriminative feature learning method have quickly become the mainstream; in recent years, with the introduction of deep learning (deep learning) method, face recognition technology has been pushed to a new level. steps. At present, the representative cutting-edge technologies in the field of face recognition are as follows: [0003] First of all, th...

AI Technical Summary

Problems solved by technology

[0005] The research group of Professor Tang Xiaoou from the Chinese University of Hong Kong conducted a more in-depth study on the face recognition technology based on deep learning. They used multiple face sub-images to train the deep neural network separately, and then connected the features output by each sub-neural network in series. A better recognition effect is obtained, but this simple concatenation of the features extracted from each sub-image loses the structural characteristics of the image itself

Images