A new approach is proposed for extraction of features from human preferences reasoning. Conditional Probability Table (CPT) is a mentality representation to control the reasoning in Bayesian Belief Network (BBN). A software tool was developed using texture analysis with a co-occurrences matrix algorithm. As a case study, it was tested on BBN of moss (Rhacomitrium canescens) produce preferences. The result successfully represented features extracted as specific patterns. It is applicable as a new computational method for reducing many concrete parameters (Dimensionality) and extracting the information from CPT in five textural features. These features are essential as abstractive parameters for designing customized agro-industrial production to provide every consumer with a produce that matches his or her unique preferences.
A rapid and simple method of quality measurement for compost was developed using an FT-NIR technique. Compost samples collected from a composting factory were used for the calibration and validation of components. NIR spectra were measured within the range of 1000-2500nm. Total carbon (TC) and total nitrogen (TN) were measured by N/C Analyzer and mineral contents measured by an Inductively Coupled Plasma (ICP) instrument for calibration and validation. Partial least square regression and multiple linear regression were employed for calibration among the second derivative spectra and measured the mineral content. The calibration models evaluated correct contents of potassium, phosphorus and other minerals of the test composts in received accuracy. These results indicated that NIR spectroscopy was a useful tool for the management of compost quality.
To ascertain the possibility of producing refuse derived fuel (RDF) using waste plastic film and sewage sludge, the current state of waste plastic film and sewage sludge generation was investigated and the heavy metal components and heating values for those materials were analyzed. At this time waste plastic film from agricultural production is being discarded in the fields and sewage sludge is being buried in landfills or discharged into the ocean, resulting in land and ocean pollution in Korea. Waste plastic film from mulching operations on farm land and sewage sludge could be used to produce RDF if they are mixed in the proper ratio to get the heating value required for standard RDF. The Hg, Cd, Pb and As content did not exceed the specified limits for RDF, demonstrating the usability of waste plastic film and sewage sludge as raw materials of RDF. From an elementary analysis of waste plastic film and sewage sludge, SO2 gas is not expected to be discharged, while NOx gas may possibly be discharged in the flue. To get the heating value required for RDF, a mixture of 50% waste plastic film and 50% sewage sludge was desirable.
Using an enzyme-linked immunosorbent assay (ELISA), a biosensor was developed to rapidly measure fungicide iprovalicarb residues in agricultural products. The biosensor was designed to include micro-pumps and solenoid valves for fluid transport, a spectrophotometer cuvette as a reaction chamber, a photodiode with a light-emitting diode for optical density measurement, and a control microcomputer to implement assay. The rate of change in optical density of the cuvette was read as final signal output. Micro-pumps were evaluated to investigate their delivery capability. The maximum error and the coefficient of variation were found to be 4.3% and 4.6%, respectively. To predict the concentration of the iprovalicarb residue in agricultural products, a linear calibration model was obtained with r-square values of 0.992 for potato and 0.985 for onion. In validation test for the samples of potatoes and onions against the high performance liquid chromatography (HPLC), very high correlation values were obtained as 0.996 and 0.993, respectively. Using the cuvette immobilized with antigen, it took 21-minutes for the biosensor to complete the measuring process of the iprovalicarb residues . The measurement assay was successfully automated with an adequate sensitivity level to measure the government limits for iprovalicarb residues.
A machine vision-based mobile eggplant grading robot was developed as a model to sense and log essential data needed for precision farming and traceability of agricultural crops. The robot consisted of battery railcar, trailer, grading mechanisms including manipulator, end-effector, machine vision and divider. Manually operated, the robot moved along crop ridges during the harvesting process with the grading mechanisms all mounted on the trailer. The robot sensed and logged the field spatial variability dynamically by grading manually harvested eggplant fruit of individual tree. This paper presents the details of manufactured robot and results of the performance test conducted. Test results showed that the robot was feasible for field spatial variability sensing in greenhouse.
Many types of agricultural facilities have been constructed at many places around the world. Although most of the operations are mechanized at these facilities, high noise from mechanical systems often prevents operators from discharging their duties efficiently and results in an unsafe environment. Virtual low-noise space is defined as the space required by operators in order to smoothly communicate in a high-noise environment, i.e., they feel as though they are working in a low-noise environment. The virtual low-noise space has three functions: to reduce the noise around the operators' ears, to communicate by speech without noise, to recreate the audio image on the basis of the operators' mutual positions. In this research, virtual low-noise space was constructed using noise-canceling headphones, microphones, a TV camera, and computers.
The publisher of "Engineering in Agriculture, Environment and Food (EAEF)" will change from J-Stage (JST) to Elsevier Ltd. on January 1, 2014. New issues after Vol.7 will be browsed on ScienceDirect, whereas Vol.1-6 can be still browsed on J-Stage.
J-STAGE will be temporary out of service due to system maintenance on Oct. 27, 22:00 - Oct. 28, 8:00 UTC.