• Proceedings of the Korean Information Science Society Conference
• /
• 2003.10b
• /
• pp.637-639
• /
• 2003

본 논문은 카메라로 획득한 배의 표면과 꼭지 영상을 입력으로 하여 RBF 신경망 기반 분류기를 사용하여 양호한 배인지 아닌지를 판별하는 판별기의 설계에 대해 설명한다. 먼저 입력 영상에서 배경을 분리시킨 후 배만을 포함하는 영상을 얻고 이 영상에서 윤곽선과 같은 여러 가지 특징들을 추출한 후 미리 대량의 표면 영상과 꼭지 영상으로 훈련시킨 두 개의 RBF 신경망 기반 분류기를 사용하여 배의 상태를 판별한다. 구현되는 세부 모듈을 과일 종류에 맞게 수정한다면 제안되는 방법을 사과, 참외와 같은 다른 과일에도 적용할 수 있을 것이다.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.7
• /
• pp.51-57
• /
• 2008

In recent computer vision or robotics fields, the research area of object recognition from image using low cost web camera or other video device is performed actively. As study for this, there are various methodologies suggested to retrieve objects in robotics and vision research areas. Also, robotics is designed and manufactured to aim at doing like human being. For instance, a person perceives apples as one see apples because of previously knowing the fact that it is apple in one's mind. Like this, robotics need to store the information of any object of what the robotics see. Therefore, in this paper, we propose an methodology that we can rapidly recognize objects which is stored in object database by using SIFT (scale invariant feature transform) algorithm to get information about the object. And then we implement the methodology to enable to recognize simultaneously multiple objects in an image.

• Journal of Biosystems Engineering
• /
• v.22 no.4
• /
• pp.433-438
• /
• 1997

The freshness of apple was evaluated by characterizing the surface texture of flesh cells. The freshness index which was related to the amount of wrinkles on the cell surface was defined to quantify the freshness. Four parameters relevant to the amount of the cell wrinkles were selected and measured using image analysis including wrinkle extraction procedure and Fast Fourier Transform of a wrinkle-extracted image. Out of 4 parameters, three parameters had highly significant correlations with the time elapsed after harvest. But it was concluded that two parameters out of such 3 parameters could be used for description of freshness index.

• Proceedings of the Korean Society for Agricultural Machinery Conference
• /
• 1999.12a
• /
• pp.575-580
• /
• 1999

• Proceedings of the Korean Society for Agricultural Machinery Conference
• /
• 1998.06b
• /
• pp.262-269
• /
• 1998

• Proceedings of the Korean Society for Agricultural Machinery Conference
• /
• 2000.02a
• /
• pp.436-444
• /
• 2000

• Proceedings of the Korean Society for Agricultural Machinery Conference
• /
• 1998.12a
• /
• pp.380-385
• /
• 1998

• Journal of Biosystems Engineering
• /
• v.24 no.6
• /
• pp.523-530
• /
• 1999

자기공명영상 (MRI)기법은 최근에 그의 이용 분야가 식품공학 및 농학 분야까지 확대되고 있다. 이 연구에서는 MRI의 기본 원리를 알아보고 어떤 요인들이 자기공명영상의 명암을 결정하는지 알아 보았다. MRI를 이용하여 비파괴적인 방법으로 농축산물의 내부 구조 영상을 얻고 이로부터 토마토의 내부 공동, 감자의 공동병, 사과와 파파야의 멍, 버찌씨의 유무, 망고와 감자의 벌레에 의한 손상, 망고와 마늘의 내부 결함, 키위의 성숙도에 따른 변화, 그리고 소고기의 내부 구조 둥을 판별할 수 있음을 알아 보았으며 농축산물의 품질을 결정짓는 이러한 내부 요인들이 어떻게 자기공명영상을 형성하는지 알아 보았다. 대상 시료에 따라 적당한 자기공명영상 인자(TR과 TE)를 선택하여야 한다. 또한 특정 인자를 강조하기 위해서는 최적의 자기공명영상 인자를 선택하여야 한다.

• Journal of Broadcast Engineering
• /
• v.25 no.2
• /
• pp.208-217
• /
• 2020

The classification of plant diseases by images captured by a camera sensor has been studied over past decades. A method that has gained much interest is to use image segmentation, from which statistical features are derived and analyzed by machine learning. Recently, deep learning has been adopted in this area. However, image segmentation is still a difficult task to achieve stable performance due to a variety of environmental variations. The end-to-end learning in neural network has a demerit that train images may be different from real images acquired in outdoor fields. To solve these problems, we propose superpixel-based disease classification method using end-to-end CNN (convolutional neural network) learning. Based on experiments performed on PlantVillage apple images, the classification accuracy is 98.29% and 92.43% for full-image and superpixel. As well, the multivariate F1-score is (0.98, 0.93). Therefore we validate that the method of using superpixel is comparable to that of full-image.

• Journal of Biosystems Engineering
• /
• v.22 no.4
• /
• pp.411-420
• /
• 1997

The mechanization efficiency using high ability machines such as tractors or combines in a paddy field rice farm is high. Mechanization in harvesting fruits and vegetables is difficult, because they are easy to be damaged. Therefore, Advanced techniques for careful handling fruits and vegetables are necessary in automation and robotization. An apple harvesting robot must have a recognition device to detect the positioning of fruit, manipulators which function like human arms, and hand to take off the fruit. This study is related to the development of a rotatic hand as the first stage in developing the apple harvesting robot. The results are summarized as follows. 1. It was found that a hand that was eccentric in rotatory motion, was better than a hand of semicircular up-and-down motion in harvesting efficiency. 2. The hand was developed to control changes in grasp forces by using tape-type switch sensor which was attatched to fingers' inside. 3. Initial finger positioning was set up to control accurate harvesting by using a tow step fingering position. 4. This study showed the possibility of apple harvesting using the developed robot hand.