WEI Yongjie, WANG Haoran, LI Hui, ZHANG Shuri

(School of Mechanical Engineering, Hebei University of Technology, Tianjin 300130, China)

Abstract：The water spray strongly absorbs and scatters light in the near-infrared band. Hyperspectral data is informative and plays an important role in water spray measurement. However, the spectral characteristics of water spray is similar to background. As a result, the principal component analysis (PCA) or minimum noise fraction algorithm (MNF) can hardly classify them respectively. A joint transformation method based on maximum containing area was proposed which was derived of the above two algorithms. Firstly, PCA was used to obtain a larger area than the true area of water spray, thus most of the background information was removed. Then MNF was adopted to obtain the real area of water spray from the results of PCA. So the water spray area was extracted. Hyperspectral images of nozzle water spray at different pressures were collected in the experiments. The results show that the area of water spray can be identified and located, the spray angle and spray area decrease as the pressure increase. It is consistent with the two-phase flow characteristics of water spray.

Keywords：hyperspectral; principal component analysis; minimum noise fraction; water spray

Get Citation：WEI Y J, WANG H R, LI H, et al. Hyperspectral water spray recognition method based on joint transform of maximum containing area[J]. China Powder Science and Technology, 2023, 29(5): 1-7.

DOI：10.13732/j.issn.1008-5548.2023.05.001

Received：2023-03-21,Revised：2023-06-01,Online：2023-07-26 17：40。

Funding Project：天津市自然科学基金项目,编号:18JCTPJC56600。

First Author：魏永杰(1971—),男,教授,博士,研究方向为光电检测和光谱分析。E-mail: yj.wei@163.com。

CLC No：TP751; TB4

Type Code：A

Serial No：1008-5548(2023)05-0001-07