The leaf chlorophyll content is one of the most important factors for the growth of winter wheat. Visual and near-infrared sensors are a quick and non-destructive testing technology for the estimation of crop leaf chlorophyll content. In this paper, a new approach is developed for leaf chlorophyll content estimation of winter wheat based on visible and near-infrared sensors. First, the sliding window smoothing (SWS) was integrated with the multiplicative scatter correction (MSC) or the standard normal variable transformation (SNV) to preprocess the reflectance spectra images of wheat leaves. Then, a model for the relationship between the leaf relative chlorophyll content and the reflectance spectra was developed using the partial least squares (PLS) and the back propagation neural network. A total of 300 samples from areas surrounding Yangling, China, were used for the experimental studies. The samples of visible and near-infrared spectroscopy at the wavelength of 450,900 nm were preprocessed using SWS, MSC and SNV. The experimental results indicate that the preprocessing using SWS and SNV and then modeling using PLS can achieve the most accurate estimation, with the correlation coefficient at 0.8492 and the root mean square error at 1.7216. Thus, the proposed approach can be widely used for winter wheat chlorophyll content analysis.

Real-world decision-making is based on human cognitive information, which is characterized by fuzziness and partial reliability. In order to better describe such information, Zadeh proposed the concept of Z-number. Ranking the Z-number is an indispensable step in solving the decision-making problem under the Z-number based information. Golden rule representative value is a new concept introduced by Yager to rank interval values. This paper expands it and proposes a new golden rule representative value for fuzzy numbers. Then apply it to the ranking of the Z-number. Some new rules involving the centroid and fuzziness of fuzzy numbers are constructed to capture the preference of decision-makers. The Takagi-Sugeno-Kang(TSK) fuzzy model is used to implement these rules. The obtained Rep function is used to construct a new golden rule representative value fuzzy subset of the Z-number and associate this new fuzzy subset with a scalar value. This fuzzy subset not only implies the fuzzy aspect of the Z-number but also contains the information in the hidden probability distribution. The scalar value is regarded as the golden rule representative value of the Z-number to participate in the ranking. The proposed method greatly retains the original information of the Z-number and can overcome the shortcomings of the existing methods. Some numerical examples are used to describe the specific process of the proposed method. The comparative analysis and discussion with existing methods clarify the advantages of the proposed method.