Image homologous identification platform based on fine registration and implementation method thereof

Abstract

The invention discloses an image homologous identification platform based on fine registration and an implementation method thereof. The image homologous identification platform based on fine registration comprises: a to-be-detected image receiving module, a digital image fine matching module, a deep learning training system module, a deep learning-based image homologous identification module andan online identification module, wherein the digital image fine matching module is connected with a first end of the image homology identification module based on deep learning; the deep learning training system module is connected with a second end of the image homology identification module based on deep learning; and the online identification module is connected with a third end of the image homology identification module based on deep learning. The image homologous identification platform based on fine registration can perform fine registration on the to-be-detected image, can automatically correct the size and rotation difference of the image formed under different conditions, can eliminate noise pollution in the image transmission process, and is good in robustness, high in responsespeed and high in system practicability.

Description

technical field [0001] The invention relates to the technical field of image analysis, in particular to an image homologous identification platform based on fine registration and an implementation method thereof. Background technique [0002] Machine learning methods are widely used in image analysis to complete specific tasks on new data, such as classification, recognition, and segmentation, by training a model on a given data set. However, traditional machine learning algorithms need to use prior knowledge to manually extract features from raw data to train the model. Due to the difficulty of feature selection, the model may have overfitting problems, and the generalization ability is difficult to guarantee; on the other hand, traditional models are difficult to adapt to large-scale data sets, and the model scalability is poor. [0003] Deep learning (Deep Learning) is the most popular research direction in the field of machine learning in recent years. Its motivation is...

AI Technical Summary

Problems solved by technology

[0008] (1) It is necessary to manually extract features suitable for digital images, but it takes a lot of time to extract suitable features for images provided by different manufacturers or commodities, and requires a lot of experience in related fields

[0009] (2) Due to the different image acquisition equipment used by users, different shooting angles, and different light conditions during shooting, the size and rotation of the generated digital images are very different, so the collected images are not standard enough, and often in the process of image transmission. There is noise pollution, and most of the current image homology identification accuracy is not high, the robustness is not good, and the practicability is not strong

The image homologous identification method of fine registration and deep learning technology has not been seen in various literatures so far

Images