The purpose of this study is to detect jamming in sealing aluminum vacuum metallized film packaging for food material. Two types of package are examined; one is unpigmented silver packaging and the other is red colored. The detection algorithm developed in this study consists of the following methods: capturing 24 bit image data of the package's seal area; decomposition of the RGB components (8 bits each); determination of the seal area, gray scale transformation, FFT, image denoising by a mask operation, IFFT, edge treatment and binary conversion by applying a brightness threshold. The optimum brightness threshold is calculated by discriminant analysis. The combined optimum brightness thresholds for the detection of jamming in unpigmented silver packaging are a brightness threshold of 55 and a threshold of 35 using the B-component of the image data. The discrimination ratio of the jamming package is 99.6% under the above conditions. For red colored packages, the optimum brightness thresholds for detecting jamming are a brightness threshold of 60 and a brightness threshold of 43 using the B-component of the image data, giving a discrimination ratio of 94.3%. It was showed that the detection of jamming in sealing food packages is possible by the detection systems developed in this study.

View full abstract

Kansei engineering aims to translate feelings and impressions into product parameters. Employing Kansei information will become increasingly important in the fields of food and the environment. This is because Kansei information will enable us to design, develop, and evaluate agricultural produce and food which are prominent targets for consumer evaluation, letting us conceive comfortable interior spaces, and create urban and rural landscapes high in amenity effect. Until now, we have investigated facial expression information and its relation to Kansei, and conducted facial expression analysis by employing the Facial Action Coding System (FACS) and facial node imaging. In this report, we developed a two-dimensional dynamic facial expression tracking system which we utilized in specifying important facial nodes. We established a Kansei inference method using facial expression information and employed degree of similarity and the nearest neighbor rule (NN rule) to conduct differentiation. The potential for facial expression - Kansei inference was suggested through differentiation by using the ratio in the change in area of a normalized polygon as facial feature vectors. We shall report on our findings here.

View full abstract

An Internet-accessible image analysis service, Remote Image Analysis Service (RIAS), was designed and implemented in this research. The Image Analysis Core, which internally carries out the actual image processing, utilizes a Java code library known as Java Advanced Imaging. Three different inter-process protocols are used for communication between RIAS and client software applications, each serving a different application category. Firstly, Java applications within a host organization's firewall communicate with the image analysis core using Java RMI, a relatively fast binary protocol. The second category of applications, namely Java clients outside of the organization, access RIAS by sending requests and receiving results through EJB Server over RMI/IIOP. These two protocols are only usable by Java clients. The third category of applications is non-Java client applications, which access RIAS through SOAP-based Imaging Service using HTTP as the transport mechanism. SOAP, as the backbone of Web Service, is a cross platform and a cross language, and hence can be accessed by any languages supporting SOAP. Imaging Service implemented in RIAS supports MTOM, the newest attachment specification. Client applications of Java applet and MS C#. Net were also developed, verifying that RIAS has good performance in three accesses. RIAS is potentially able to be used by mobile devices such as cell phone with camera and PDA with camera. This is important because mobile devices usually lack of ability of floating point computing, by using RIAS, agricultural researchers and farmers may carry out image processing in the field.

View full abstract

In response to frequent heavy rains and pond-floods recently, we have developed a disaster prevention information system to reduce flood damage to nearby residents and surrounding lands caused by irrigation pond overflows. The system utilizes a database of all ponds in each prefecture together with rain prediction data from each city accessed via the Internet. In the case of some high-risk ponds in a small watershed, the system automatically sends alerts of emergency- or flood-related information to PC's and cellular phones using GIS images with voice and E-mail, respectively. We developed the GIS on Microsoft Excel using MapWindow^TM GIS for ease of PC environment accessibility. Two case studies, a simulation using Fukui heavy rain data and in-situ observation during some heavy rains, made it clear that the system predicts the dangers of flooding and variations in water level. For practical use, this system needs an inexpensive instrument for measuring initial pond water level and transmitting it to the main system. Such an instrument is now being developed.

View full abstract

In the farm management field, weather variance is a significant risk factor. Comprehensive techniques to assist in the management of wide-ranging weather risks are indispensable.
In this paper, basic technologies for assessing climatic risks are introduced, and their suitability for farm management is assessed. The new technologies introduced in the paper have a great deal of potential because they can make amends for all of the variance risk inherent in farm management regardless of the damage scale of climatic damage.
In order to integrate the new technologies into farm management, we are advancing improvement in the platform.

View full abstract