A Classification Method of Submarine Substrate Sonar Image Based on Convolutional Neural Network

Abstract

The invention discloses a method for classifying sonar images of sea bottom quality based on a convolutional neural network, which belongs to the technical field of image classification. Specifically, to obtain the sonar image of the seabed bottom, denoise and enhance the image, extract the edge shape based on the Canny algorithm, generate a gray-element co-occurrence matrix, construct a convolutional neural network classifier structure and sample set, and train the neural network. Network to obtain classification models and implement sonar image classification of seabed bottom. The present invention focuses on the graphics characteristics of sonar images of seabed bottom, solves the shortcomings of using a single method, and uses the learning strategy of the convolutional neural network classifier structure to learn and train different types of seabed bottom, and finally obtains a classification The function classification model achieves the purpose of fast and accurate classification of seabed bottom sonar images.

Description

technical field [0001] The invention belongs to the technical field of image classification, and in particular relates to a method for classifying sonar images of sea bottom quality based on convolutional neural networks. Background technique [0002] With the rapid development of sonar technology, sonar images of seabed bottom can contain relatively rich information about seabed topography and bottom characteristics. Protection is of great significance, and the classification of sonar images of seabed bottom has also become a research hotspot. However, limited by the underwater complex sound field environment and the performance of sonar equipment, sonar images have problems such as serious speckle noise interference, blurred edge features, low contrast, and uneven brightness. Issues that require urgent attention. [0003] Now widely used classification methods can be divided into two categories: unsupervised classification and supervised classification according to the c...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problems of weak generalization ability, long training time, and high requirements on the quality of training samples in the prior art in the classification of sea bottom sonar images, and propose a sea bottom based on convolutional neural network. Quality sonar image classification method, described method specifically comprises steps:

Images