Underwater laser multi-target end-to-end automatic recognition system based on swarm intelligence optimization

Abstract

The invention discloses a swarm intelligence optimized underwater laser multi-target end-to-end automatic recognition system, which comprises an underwater target laser image collection module, a swarm intelligence optimized underwater laser multi-target end-to-end automatic recognition system and a display module, the swarm intelligence optimization underwater laser multi-target end-to-end automatic recognition system comprises an underwater laser image database, a preprocessing module, an underwater laser multi-target modeling module, a swarm intelligence optimization module, an underwater laser multi-target automatic recognition module and an underwater target recognition output module. According to the method, optimal end-to-end automatic model training and target recognition are realized, multiple targets in the same underwater laser image can be recognized, the recognition accuracy is high. Meanwhile, the method has the advantages of high feature extraction capability, high recognition speed, high intelligent degree and the like, and the problems that traditional underwater laser target recognition is tedious in operation and high in efficiency are solved. The method is only suitable for a single target, is low in recognition accuracy, is low in speed, is poor in model quality, and needs to manually select parameters.

Description

technical field [0001] The invention relates to the field of underwater laser target recognition, in particular to an underwater laser multi-target end-to-end automatic recognition system optimized by swarm intelligence. Background technique [0002] The ocean area accounts for about 70% of the earth's surface. The rich minerals, organisms, energy and other resources in the ocean are huge wealth endowed by nature to human beings. After the blue-green light attenuation of blue-green laser is weaker than that of other bands, the detection application of blue-green laser in seawater has become the focus of many scholars. At the same time, underwater detection based on acoustic waves is seriously affected by seawater, terrain, and marine organisms, which seriously affects the quality of sonar underwater detection, while underwater laser detection is not easily affected by seawater temperature, salinity, etc., and can be directly imaged in two dimensions. Better target ranging a...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the current underwater laser target recognition problems of low accuracy, slow speed, poor model quality, manual selection of parameters, and only simple target recognition, the purpose of the present invention is to provide a high accuracy, fast recognition speed, intelligent It is a system with a high degree of automation that can be used for end-to-end automatic recognition of underwater laser multi-targets. Based on underwater laser images, an intelligently optimized end-to-end automatic recognition system for underwater targets is constructed to achieve optimal underwater laser multi-target recognition. The end-to-end automatic recognition system for underwater laser multi-targets optimized by swarm intelligence uses laser images containing one or more underwater targets as input, and improves the quality of underwater laser images through image enhancement. Underwater laser multi-target recognition model modeling, swarm intelligence optimization module is used to optimize the model to obtain the optimal recognition model, and use the constructed underwater laser multi-target automatic recognition model to carry out newly acquired underwater laser target recognition, and finally output the recognition results

The invention realizes optimal end-to-end automatic model training and target recognition, can identify multiple targets in the same underwater laser image, has high recognition accuracy, and has the advantages of strong feature extraction ability, fast recognition speed, and high degree of intelligence. , to solve the traditional underwater laser target recognition needs target segmentation, feature extraction, classifier modeling and other cumbersome steps, only suitable for single target, low recognition accuracy, slow speed, poor model quality, manual selection of parameters and other shortcomings

Images