A high-resolution image classification method and classification device

Abstract

The invention discloses a classification method and a classification device for high-resolution images, which are used to improve the classification accuracy of high-resolution images. The classification method includes: dividing the high-resolution image into several sub-images of preset sizes, and making overlapping regions exist between at least two adjacent sub-images; inputting each of the sub-images into a convolutional neural network wherein each of the sub-images is correspondingly generated with multiple sub-feature maps; the multiple sub-feature maps generated corresponding to each of the sub-images are placed in a preset order to form a feature map to be classified; The feature map to be classified is classified through the fully connected layer in the neural network, and the classification result is output.

Description

technical field [0001] The invention relates to the technical fields of image classification and deep learning, in particular to a classification method and a classification device for high-resolution images. Background technique [0002] The concept of deep learning originated from the research of artificial neural networks, including multi-layer perceptrons with multiple hidden layers. Deep learning combines low-level features to form more abstract high-level representation attribute categories or features to discover the distributed feature representation of data. At present, deep learning has a wide range of applications in academia and industry. [0003] Convolution Neural Networks (CNN) is a feed-forward neural network whose artificial neurons can respond to surrounding units within a part of the coverage, including convolutional and pooling layers. For a long time, although CNN achieved the best results in the world at the time on small-scale problems, such as handwr...

AI Technical Summary

Problems solved by technology

[0005] In recent years, with the advent of the era of big data, deep learning technology has been applied in various fields, and its application scenarios have been extended. However, the application of deep learning on images is still limited to the input of small-sized images. For high-resolution High-rate medical images, the current general solution is to first compress it into a smaller image, then input the smaller image into CNN for processing, and then classify the image after processing; however, the existing technology will make the compressed image Images are distorted, making it difficult to accurately distinguish image categories

Images