Random convolutional neural network-based high-resolution image scene classification method

Abstract

The invention discloses a random convolutional neural network-based high-resolution image scene classification method. The method comprises the steps of performing data mean removal, and obtaining a to-be-classified image set and a training image set; randomly initializing a parameter library of model sharing; calculating negative gradient directions of the to-be-classified image set and the training image set; training a basic convolutional neural network model, and training a weight of the basic convolutional neural network model; predicting an updating function, and obtaining an addition model; and when an iteration reaches a maximum training frequency, identifying the to-be-classified image set by utilizing the addition model. According to the method, features are hierarchically learned by using a deep convolutional network, and model aggregation learning is carried out by utilizing a gradient upgrading method, so that the problem that a single model easily falls into a local optimal solution is solved and the network generalization capability is improved; and in a model training process, a random parameter sharing mechanism is added, so that the model training efficiency is improved, the features can be hierarchically learned with reasonable time cost, and the learned features have better robustness in scene identification.

Description

technical field [0001] The invention belongs to the field of remote sensing image processing, and in particular relates to a high-resolution image scene classification method based on a random convolutional neural network. Background technique [0002] High-resolution remote sensing images (generally referring to remote sensing images with a spatial resolution greater than 1 meter) have been widely used in various fields because of their high spatial resolution and rich spatial information that can provide fine information for object recognition. However, due to the relatively high spatial resolution of the image, the scene for recognition usually contains pixels obtained by mixing a variety of different types of ground objects. These ground objects often have different structural information, but due to the low spectral resolution, it is often difficult to distinguish . With the maturity of high-resolution imaging technology and the reduction of cost, high-resolution image...

AI Technical Summary

Problems solved by technology

[0007] The present invention mainly provides a stochastic convolutional neural network model based on gradient boosting to solve the problems existing in existing methods, and provides a method that can not only extract high-level features hierarchically, but also self-adaptively aggregate multiple network models. method

Images