Short-wave Infrared Spectroscopy Analysis Of Typical Aquatic Target Materials And The Characteristic Band Selection

Global economic integration has promoted the rapid development of water transportation industry.The water transportation environment is complex and changeable,and shipwreck accidents such as ship collision and self-sinking occur from time to time.When carrying out water emergency rescue tasks,various water targets are usually floating on the surface,and their morphological characteristics are not obvious.The reflection brightness difference between water targets in ordinary visible light and RGB image is small,and water targets are vulnerable to light,and it is difficult to identify.This paper introduces short-wave infrared（SWIR）spectral imaging technology to study and analyze the SWIR spectral characteristics of aquatic target materials,and selects the characteristic band of aquatic target based on the SWIR spectral image characteristics.These images of SWIR characteristic bands can effectively suppress water surface light clutter and lay a foundation for further water target identification search and rescue.The main research contents of this article are as follows:（1）The reflection spectral characteristics of four kinds of water target materials such as nylon life jacket,wood fiber board,polystyrene foam plate and aluminum alloy plate were analyzed.The SWIR multispectral imaging system was built independently to collect the SWIR multispectral images（950--1700 nm）and analyze the spectral characteristics of the water target materials.The experimental results show that: in the short-wave infrared band,the spectral characteristics of the four types of water target materials are very different,and very different from the water background,which is conducive to inhibit the background and highlight the water target.（2）According to the requirements of water target identification for target imaging,based on the spectral characteristics of the four types of water target materials,three band selection methods are information comparison method,gray value comparison and entropy method,and data fusion decision are adopted to select SWIR characteristic bands for the four types of water targets.Through experimental data analysis,1080 nm,1200 nm,1250 nm,1290 nm and 1550 nm were selected as characteristic bands.These five characteristic bands show the effective SWIR spectral characteristics of the water target material,and show the bright image characteristics in the SWIR images under the five characteristic bands,providing a theoretical basis for the rapid and effective identification of water targets.（3）For the selected SWIR feature band images,grayscale adaptive classification and HOG + SVM classification are used to identify the feature band image targets,and evaluate the effectiveness of the five feature bands by using the target recognition performance index.Two target recognition performance indexes of recognition rate and false alarm rate were selected to compare the imaging performance of five characteristic bands and five other bands.Experiments show the better identification effect of target identification application on characteristic band launching.Further combine the selected SWIR characteristic band,and improve the aquatic target SWIR imaging device prototype based on automatic filter switching.