Research On Estimation Of Wheat Lai Based On Case-based Reasoning

Precision Agriculture is management and business philosophy of modernagriculture developed in the1990s based on information and knowledge, has becomean important form of modern agricultural production of rational utilization ofagricultural resources, improving agricultural crop production, reducing theproduction cost and improving the ecological environment. Getting information onfield crops by dynamic monitoring of crop growth could help to keep abreast of cropgrowth conditions, growth of the seedlings, nutritional status and its changes, it’s easyto take a variety of management measures to ensure the normal growth of crops.However, due to the lack of low-cost, high-density, high-precision and high reliabilityand efficient way of getting agricultural information, quick access to field cropsinformation has been the "bottleneck" of development of precision agriculturetechnology. With the promotion and application of hyperspectral remote sensingtechnology, it bring new opportunities to solve this problem. Carrying out research ofleaf area index (LAI) estimation using hyperspectral has become one of the hot issuesin the field of access to information in precision agriculture.In this paper, spectral reflectance data measured using ASD portablespectrometer and wheat LAI measured in synchronization laboratory in an area ofwinter wheat in Beijing as the data source including three major wheat growingperiod in2010and2012, and conducted a systematic study to estimate wheat LAI.The main work done is as follows:Firstly, the method of building statistical models was used to estimate LAI ofwheat. On the basis of wheat canopy spectral reflectance data, selected thecharacteristic bands and then calculated10vegetation indices. In each growth period,four kinds of statistical regression models including linear model, exponential model,logarithmic model and polynomial model were constructed with LAI as the dependentvariable and vegetation index as the independent variables for each vegetation index,and then verified estimation results.Secondly, the approach of case-based reasoning was introduced to estimatewheat LAI, studied and implemented the key technologies of estimation of wheat LAIbased on case-based reasoning. Used tuple to represent the case of wheat; Employedthe retrieval of templates and K nearest neighbor as case retrieval strategies based on the characteristics of the wheat case; Kolodner distance formula was used to calculatethe similarity; Case reusing achieved by using the weighted average method; Utilizedthe genetic algorithm to optimize the weights of case attributes. The method ofcase-based reasoning for estimating wheat LAI has higher estimation accuracyverified by experiment.At the end, the software platform of wheat agronomy parameter estimation basedon intelligent computing was developed and achieved a simple, convenient way toestimate wheat LAI. At the same time, the platform laid a foundation on continuoingto carry out applied research in wheat agronomy parameter estimation based oncase-based reasoning.