Critical Review of Fast Detection of Crop Nutrient and Physiological Information with Spectral and Imaging Technology
He Yong,Peng Jiyu,Liu Fei,Zhang Chu,Kong Wenwen
DOI: https://doi.org/10.3969/j.issn.1002-6819.2015.03.024
2015-01-01
Abstract:The research achievements and growing trends of spectral and imaging technology in fast detection of crop nutrient and physiological information were reviewed. Firstly, the principle of spectral and imaging technology, the data processing methods, modeling methods and the indexes of model evaluation were briefly introduced in this paper. Secondly, this paper focused on the research achievements and applications of spectral and imaging technology in fast detection of crop nutrient and physiological information of five kinds of crops (i.e. rice, wheat, oilseed rape, maize, soybean), including chlorophyll content and nitrogen content detection, crop diseases and pests monitoring, stress diagnosis (water, heavy metal, weed, pesticide stress) and yield prediction. In nutrient content and chlorophyll content detection, the data was acquired by ground-based sensing, aircraft-based sensing and satellite-based sensing, and the raw spectra, as well as vegetable indices, were used to build quantitative models. In crop diseases and pests monitoring, spectral and imaging technology were used to discriminate the crop diseases and pests, and diagnose the crop stress level. As for stress diagnosis, several recently-reported researches were briefly reviewed. In yield prediction, this paper was mainly focused on predicting the canopy parameters which were found to be significantly related to crop yield. Although the ability of spectral and imaging technology was proved, there were several problems needed to be solved. 1) The detection of crop nutrient and physiological information with spectral and imaging technology is affected by crop type, crop growing stage, operational conditions, environmental parameters and field management. Therefore, the stability and reliability of the model needs to be improved, which can be overcome by choosing suitable pretreatment methods and chemometrics methods or proposing new vegetable indices which are insensitive to these influencing factors. 2) Multi-scale rapid detection of crop nutrient and physiological information is required in the future, including the multi-scale dataset fusion, the research of different-scale sensing effect. 3) The quantitative models for the level of crop stress diagnosis is hard to carry out, due to the lack of stress assessment indexes. So it is critical to set up the reference principle for the crop stress level. Furthermore, this paper also analyzed the growing trends of the spectral and imaging technology for the fast detection of crop nutrient and physiological information. Firstly, there is a need to develop more stable and reliable methods for variable selection, data mining and model calibration, as well as the calibration technology which is based on the actual physical model between the radiation and the crop tissue. Secondly, the development of portable machinery and the online detection system for crop information acquirement is required in the further study. Likewise, further research is necessary with respect to developing the information detection system of whole crop growing stage with consideration of different crop features. In conclusion, it is proved that spectral and imaging technology can be used to detect the crop physiological information, carry out the online monitoring of the crop growing statue and the response to the adversity and stress, which is important for the realization of precision, digitization, informatization and the intelligentization management of the agriculture.