Optical sensing in agriculture is very powerful because it gets all information of agricultural products on physical and chemical level. In this paper, however, we deal with only mid-infrared (MIR) spectroscopy. Theoretically speaking, mid-infrared spectroscopy has more advantages on the spectral information than near-infrared one, while it has been very difficult to extract the solute spectrum from the water solution such as foods and agricultural products of high moisture content because of the strong absorption of MIR radiation by water. In parallel with the development of Fourier transform infrared (FT-IR) technology, various sampling techniques: attenuated total reflection (ATR), diffuse reflection, and photoacoustic methods, have been developed. Especially the FT-IR/ATR method has substantial potential as a qualitative and quantitative analytical tool for food processing. In this review, recent developments in mid-infrared spectroscopy, focusing on FT-IR/ATR method, have been taken. As an overview of the spectroscopic research on sugars, fundamental research on both the qualitative and quantitative analysis of foods and agricultural products has been reviewed. In addition, this paper also reviews some applications of FT-IR/ATR methods to monitoring metabolism during cultivation process. These works shows an important first step toward the development of nonintensive on-line monitoring devices for food processing and bioprocesses. Furthermore, we survey the importance of infrared spectroscopic characteristics of foods in order to evaluate the effects when infrared irradiation is applied to the thermal operations of food process such as heating and drying.

View full abstract

The Taguchi Method, one of the methods of robust engineering, has been a leading tool in quality engineering because the S/N ratio as utilized in this method represents the factorial effects of control variables to outputs while also taking noise factors into consideration. In this method, it is said that control variables with high interactions should not be included in design parameters. Therefore it is important to select control variables without high interactions. It is almost impossible, by using conventional experimental methods, to identify these interactions for every combination of control variables. Recently, Artificial Neural Network (ANN) has been utilized to describe various non-linear systems such as the performance of combine harvesters including the separating and cleaning functions etc. In this paper, ANN was used to model these non-linear systems of combine harvesters in order to identify the hidden interactions between the control variables. The identification made it possible to select the proper control variables. After the selection of the proper control variables, the factorial effects of the control variables were estimated by a developed ANN model and by experimental results based on an L18 matrix. The factorial effects estimated by the developed ANN model were more accurate than those estimated by experimental results based on the L18 matrix. The obtained factorial effects of the control variables facilitate designing the separating and cleaning functions of combine harvesters. It was proved that the proposed method combining the Taguchi Method and ANN is effective.

View full abstract

A system for image measurement of shape information of leaves, named "RTL: RealTime Leaves", was developed. RTL is the first system to implement the image measuring method proposed by Shono (1995) and Shono et al. (1996) for extracting 3-dimensional shape information from multiple leaf images. RTL can, via Internet, use images saved automatically and periodically on distant Web-servers, and can extract time series of growth status indices using built-in neural networks.
RTL consists of several components as follows; (1)an image data logger, (2)a texture analyzer, (3)a neural network for calculating shape information, (4)a procedure for reconstructing 3-dimensional shape information, (5)a neural network for monitoring growth status indices. Furthermore, an external supporting subsystem for the construction of neural networks maintains the effectiveness of the neural networks against changes in leaf shape with growth. This subsystem can be used to interactively and easily define and train up the neural networks.
RTL is designed to function as a reliable doctor, who cautiously monitors plants' growth status indices to find signs of undesirable growth status, for example wilting. After finding such a sign, RTL can not only show a warning message, but also show a realtime plant image. This image is exaggeratedly deformed to reflect the recent index fluctuations, allowing easy and quick understanding of the situation.
From the results of an experimental trial using remote images of tomato plants, it was shown that RTL could complete its image measurement processes within about 6 minutes, and could properly realize its intended functions.

View full abstract

A Java class library, called "Geostatistics Library" was developed for evaluating field data using Geostatistics. Numerical calculating and decision processing of field data consisting of coordinates and values are carried out within the library and the results displayed as an output map. The library was linked into java application and evaluated. The evaluation results showed that the output maps calculated by the library using semivariance and kriging interpolation were similar to those calculated by commercial software package GS+. The library was proved sufficiently useful for evaluating precision farming field data.

View full abstract

This paper proposes a method to communicate video images for diagnosis of plant diseases and pests on the Internet. Images for diagnosis have two types of region. They are a region of interest (ROI) and a region of no interest (RONI). In the proposed method, images are divided into a region of interest and a region of no interest. Then, images are layered by spatial frequency and coded. Priorities are given to packets according to region and layer. Routers transmit packets according to priority and the congestion degree of the network. Using the proposed method, we can transmit images that are best for the available network. The policy to prioritize packets is as follows. The ROI is given top priority. For the RONI, the low spatial frequency layer is given higher priority and the high spatial frequency layer is given lower priority. If severe network congestion occurs, routers begin to discard packets that have low priority in RONI. If congestion occurs harder, routers discard packets from the RONI that have high priority. Using the proposed method, we can transmit images that are best for the available network bandwidth. From the experimental result, we verified the effectiveness of the proposed method.

View full abstract

Recently one of the major challenges in agriculture has been to optimally plan fertilizer applications, maintaining proper soil fertility and reducing environment load by precisely managing the farm. In many research investigations, analyses of spatial and temporal variability in soil and yield data are performed with the help of commercial GIS software and numerical calculation software. However, it is difficult to manage and share such data and also to introduce and maintain the software, which requires ongoing investment. In this research, a Web-based system has been developed to solve these problems for precision farming research. The title of the system that has been developed is "Web-Based Precision Farming Support System". This system is intended to be used by precision farming research groups and allows centralized management, data sharing between users, and evaluation of precision farming data. In this research, the developed method adopted was to assemble software libraries of functions to facilitate their reuse within in-house developed software applications.

View full abstract

[in Japanese]

View full abstract