Abstract
Human health and ecological environment are badly affected due to oil pollution. A novel strategy for identifying oil pollutants has been proposed based on generic modular design. A total of three modules are included in the oil identification strategy and each module is consisted of a series of steps. The different steps of each module were experimented and evaluated by excitation-emission matrix fluorescence spectroscopy data set of oil. The experimental results show that the average accuracy with 13.6% was improved by using the histogram equalization than using thresholding in module 1. The average accuracy with 5% was improved by using the low-order Zernike moments than using high-order Zernike moments in module 2. The average accuracy with 28.9% was improved by using angle similarity measure in the nearest-neighbor classifier compared to the other six in module 3. The optimal accuracy with 95% was obtained by combining the margin features of excitation-emission matrix fluorescence spectroscopy extracted by low-order Zernike moments with the nearest-neighbor classifier applied to angle similarity measure. The combination also has a good specificity and sensitivity. The results provide references for identifying oil pollutants.
Highlights
Oil products are one of the important energies and chemical raw materials and may be released into the environment during exploitation, transportation, storage and use [1]
The low-order ZMs extracted from binary images (LSF) was obtained by selecting the thresholding in the third step of module 1 and the low-order Zernike moments (ZMs) in module 2
In this paper, a novel strategy has been introduced for identifying oil pollutants
Summary
Oil products are one of the important energies and chemical raw materials and may be released into the environment during exploitation, transportation, storage and use [1]. Y. Cui et al.: Novel Strategy for Identifying Oil Pollutants Based on EEMF Spectroscopy and ZMs combined data features extracted using data decomposition or unfolding algorithms with multivariate classification tools. Image processing algorithms was widely employed to extract features from EEMF for the classification [14]–[16], and have proven to be advantageous in three-way data analysis [17]. Image processing algorithms such as two-dimensional linear discriminant analysis (2D-LDA) [14] and two-dimensional principal component analysis (2D-PCA) [17] directly perform discriminant feature analysis on image matrices rather than vectors.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
