|
The outbreak of red tide seriously affects marine ecology and exploitation of fishery resources, so it is necessary to monitor for suspended particles in seawater. Due to the characteristics of various types and great changes of suspended particles in seawater, a detailed classification method based on big data is needed, and polarized light technology has great potential in this respect. In this study, we have designed a prototype for measuring absorption and scattering properties based on polarized light illumination by adding polarization state generation module and polarization state analysis module on the base of a commercial instrument named AC-S. The prototype can measure the intrinsic optical properties of water for different incident polarized light, including water extinction coefficient c(λ) and absorption coefficient a(λ). In addition, the prototype can also measure the polarization scattering information of suspended particles, which is closely related to the complex refractive index, morphology and microstructure of particles. The polarization properties of water bodies are represented by Stokes vectors. The instrument is illuminated by LED with a central wavelength of 532 nm. During the measurement, a pump drives the sample through the flow tube for detection. The polarization generation module produces a specific incident polarization beam that is directed through an optical window into the flow tube. The light signal, which is absorbed and scattered by the suspended particles in the flow tube, is then received by the polarization analysis module, which completes the measurement of light intensity and polarization. The experimental results show that for the same sample, the inherent optical properties are different under different incident polarization states, which is closely related to the properties of particles in water. We have classified the polarization data of water bodies containing different particles with the help of support vector machine (SVM) algorithm, and all of them have obtained more than 90% classification accuracy.
|
