Korea is the second largest coffee consuming country in Asia after the Philippines. For modern people, coffee has jumped over a favorite food and grown into a single culture. There are many processing ways to make coffee. In this study, we utilized magnetic resonance imaging(MRI) based on the principles of nuclear magnetic resonance(NMR) and achieved to acquire images with an non-destructive and non-invasive way. The samples we used in the experiment were ‘Robusta’ coffee bean(Congo). Magnetic resonance(MR) image sets were acquired using a MRI system, installed at Institution for Agricultural Machinery & ICT Convergence at Chonbuk National University. From the raw phase(Green Bean stage) to the roasted phase, we obtained MR images from each phase to monitor the internal changes. We divided experiment into 5 stages, starting with ‘Green Bean’ phase to ‘Roasted’ phase. We obtained images every 6 hours during the ‘Fermentation’ phase and every 3 hours during the ‘Dry’ phase. In MRI, we used a gradient echo pulse sequence to scan fast and to take images right after each experiment stage ends. The direction of imaging plane was coronal 30 images with 64 mm x 64 mm field of view(FOV). As MRI uses the magnetic properties of nuclei which especially hydrogen nuclei from water molecules, images could see clearly with sufficient moisture, but in ‘Dry’ phase, images obtained with noise involved. These result suggested that MRI technique was an efficient method to monitor the moisture distribution changes inside the coffee beans.

Numerous experiments have demonstrated that infrared thermographic methods are effective for detection of subsurface defects in the materials. The response of the material to the thermal stimulus is dependent on the existence of subsurface defects and their features. In order to obtain the information about defects, the material’s response to the thermal stimulus is studied. In this study, image processing was applied to infrared thermography images to detect defects in metals that were widely used in industrial fields. When analyzing experimental data from infrared thermographic testing, thermal images were often not appropriate. Thus, four point method was used for processing of every pixel of thermal images using MATLAB program for quantitative evaluation of defect detection and characterization which increased the infrared non-destructive testing capabilities since subtle defects signature became apparent..

본 연구는 식물체 캘러스의 접종공정의 자동화기술 개발하기 위하여 캘러스를 인식하는 영상처리 시스템과 캘러스를 새로운 배지로 접종하는 접종용 엔드이펙터, 매니퓰레이터 및 자동화 제어장치를 개발하고 그 성능을 평가하기 위하여 수행되었으며, 본 보에서는 영상처리 시스템 및 매니퓰레이터에 대한 연구결과를 소개한다. 캘러스의 인식을 위하여 영상처리부와 조명부로 구성한 영상처리 장치를 개발하였다. 영상처리부는 CCD 카메라와 PC로 구성하였으며, 조명부는 55W/3 파장 램프를 전면과 측면에 각각 설치하여 항상 일정한 조도가 유지되도록 하였으며, R, G, B 각 프레임별 화소특성을 검토한 결과, 캘러스와 배양용기는 B 프레임에서 분리가 가능하였다. 또한 캘러스 분할을 위한 알고리즘의 개발결과 캘러스가 원형인 경우는 도심을 기준으로 절단하고 장변형인 경우는 길이방향의 수직으로 절단하도록 하였으며, 장변형의 경우 위치하고 있는 방향을 인식하여 0~180˚ 범위내의 결과 값을 로봇측에 전송하여 엔드이펙터의 방향을 결정하도록 하였다. 영상취득부터 캘러스의 분할위치 결정까지 1사이클에 소요되는 시간은 디스플레이 1.5초, 영상처리 0.7초 등으로 총2.2초였으며, 개발된 인식 알고리즘의 정확도는 용기인식의 경우 전체 30개중에서 28개가 정확히 인식되어 성공률은 93%가 되었다.