Research On Image Search Technology Based On Deep Learning Neural Network

Searching for similar images in large-scale image data sets is a hot research topic in recent years,and it has been widely used in all aspects of life.The image search technology extracts a series of images similar to or identical to the image to be queried in the image database we created.As for the similarity between images,it is necessary to judge by the features between the images.Most of the current image retrieval is Content-based Image Retrieval(CBIR),compared to its previous text-based image search.This method mainly involves analyzing the image itself including the size,shape and color.The content information of the other aspects,and further search for similar images from the image library based on the image information.However,due to the increasing scale of data in today's society,there is an increasing demand for image search systems in terms of data storage,processing speed and accuracy of results.Deep Learning is a relatively dynamic machine learning method that can automatically perform task learning in recent years.It processes text,sound,and images by mimicking the mechanism by which the human brain processes information.It is a deep neural network(DNN)with multiple hidden layers connected by weights.It can learn a large amount of data to obtain useful data features,and improve the accuracy of the model for data prediction and classification.Compared with shallow neural networks,deep learning usually contains more than one layer of hidden layers.The increase of network depth is beneficial to learn more data characteristics from the big data to describe the inherent information of the data.This paper first briefly introduces the advantages and disadvantages of text-based and content-based image search technology,and then introduces the relevant knowledge of deep learning and related techniques and algorithms of convolutional neural networks,focusing on inverted index and elevant knowledge of hash coding.In this paper,an image search system based on deep learning is designed based on inverted index and hash coding.Compared with the traditional image search system,the system does not need to manually extract theexpression features of the image,and can directly learn a large number of original images,which eliminates the inaccuracy of manual annotation and can also significantly improve the image search efficiency.