Convolutional neural network picture feature extraction-based ship type identification method

Abstract

A convolutional neural network picture feature extraction-based ship type identification method disclosed by the present invention comprises the following steps of 1) acquiring a lot of unclassified water running ship pictures to carry out the image pre-processing, taking the processed pictures as a convolutional neural network sparse self-coding training data set to train to obtain a ship type feature extraction convolutional neural network; 2) extracting the training set from the water running pictures, marking the types of the ship pictures, and then inputting in the trained convolutional neural network to extract the picture features of the pictures as the training data set of a Softmax classifier to train; 3) inputting the ship pictures needing to be identified in the convolutional neural network to extract the ship features in the pictures, and inputting the ship features in the trained Softmax classifier to classify and outputting the ship types corresponding to the ships. The convolutional neural network picture feature extraction-based ship type identification method of the present invention can apply to the continuous changing environments and the undetermined inputted data, and can identify the ship types rapidly and accurately.

Description

technical field [0001] The invention relates to intelligent transportation technology, in particular to a ship type recognition method based on convolutional neural network picture feature extraction. Background technique [0002] With the development of modern transportation, intelligent transportation system has become one of the most important research fields. Intelligent transportation aims to monitor and feed back traffic flow and sudden traffic accidents. Therefore, ship type recognition has become the most advanced research direction of intelligent transportation. [0003] However, with the development of water transportation, ship monitoring has also become an increasingly serious regulatory issue. Ship monitoring mainly includes operations such as voyage records and ship type records of ships in the watershed through video monitoring equipment such as electronic eyes. According to the statistics of relevant departments, my country currently has 172,000 ships operat...

AI Technical Summary

Problems solved by technology

However, the system is difficult to install and prone to failure. It only stays in the theoretical experiment stage and is difficult to promote in practical applications.

Images