• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.40 no.3
• /
• pp.305-315
• /
• 2016

This study aims to develop a robot vision control scheme using the Newton-Raphson (N-R) method for the uncertainty of circumstance caused by the appearance of obstacles during robot movement. The vision system model used for this study involves six camera parameters (C1-C6). First, the estimation scheme for the six camera parameters is developed. Then, based on the six estimated parameters for three of the cameras, a scheme for the robot's joint angles is developed for the placement of a slender bar. For the placement of a slender bar for the uncertainty of circumstances, in particular, the discontinuous robot trajectory caused by obstacles is divided into three obstacle regions: the beginning region, middle region, and near-target region. Then, the effects of obstacles while using the proposed robot vision control scheme are investigated in each obstacle region by performing experiments with the placement of the slender bar.

• Journal of Korea Society of Industrial Information Systems
• /
• v.16 no.2
• /
• pp.65-72
• /
• 2011

In this paper, a real-time monocular vision based navigation system is developed for the disabled people, where online background learning and vector field histogram are used for identifying obstacles and recognizing avoidable paths. The proposed system is performed by three steps: obstacle classification, occupancy grid map generation and VFH-based path recommendation. Firstly, the obstacles are discriminated from images by subtracting with background model which is learned in real time. Thereafter, based on the classification results, an occupancy map sized at $32{\times}24$ is produced, each cell of which represents its own risk by 10 gray levels. Finally, the polar histogram is drawn from the occupancy map, then the sectors corresponding to the valley are chosen as safe paths. To assess the effectiveness of the proposed system, it was tested with a variety of obstacles at indoors and outdoors, then it showed the a'ccuracy of 88%. Moreover, it showed the superior performance when comparing with sensor based navigation systems, which proved the feasibility of the proposed system in using assistive devices of disabled people.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.18 no.6
• /
• pp.123-130
• /
• 2014

In this study, a vision-based displacement measurement system is developed to accurately measure the displacement of a structure with locating the camera at arbitrary position. The previous vision-based system brings error when the optical axis of a camera has an angle with the measured structure, which limits the applicability at large structures. The developed system measures displacement by processing the images of a target plate that is attached on the measured position of a structure. To measure displacement regardless of the angle between the optical axis of the camera and the target plate, planar homography is employed to match two planes in image and world coordinate systems. To validate the performance of the present system, a laboratory test is carried out using a small 2-story shear building model. The result shows that the present system measures accurate displacement of the structure even with a camera significantly angled with the target plate.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.29 no.4
• /
• pp.301-308
• /
• 2016

In this study, a novel method referred as non-marker vision-based displacement measuring system(NVDMS) was introduced in order to measure the displacement of structure. There are two distinct differences between proposed NVDMS and existing vision-based displacement measuring system(VDMS). First, the NVDMS extracts the pixel coordinates of the structure using a feature point not a marker. Second, in the NVDMS, the scaling factor in order to convert the coordinates of a feature points from pixel value to physical value can be calculated by using the external conditions between the camera and the structure, which are distance, angle, and focal length, while the scaling factor for VDMS can be calculated by using the geometry of marker. The free vibration test using the three-stories scale model was conducted in order to analyze the reliability of the displacement data obtained from the NVDMS by comparing the reference data obtained from laser displacement sensor(LDS), and the measurement of dynamic characteristics was proceed using the displacement data. The NVDMS can accurately measure the dynamic displacement of the structure without the marker, and the high reliability of the dynamic characteristics obtained from the NVDMS are secured.

• Smart Media Journal
• /
• v.11 no.4
• /
• pp.30-37
• /
• 2022

Disease threatens plant growth and recognizing the type of disease is essential to making a remedy. In recent years, deep learning has witnessed a significant improvement for this task, however, a large volume of labeled images is one of the requirements to get decent performance. But annotated images are difficult and expensive to obtain in the agricultural field. Therefore, designing an efficient and effective strategy is one of the challenges in this area with few labeled data. Transfer learning, assuming taking knowledge from a source domain to a target domain, is borrowed to address this issue and observed comparable results. However, current transfer learning strategies can be regarded as a supervised method as it hypothesizes that there are many labeled images in a source domain. In contrast, unsupervised transfer learning, using only images in a source domain, gives more convenience as collecting images is much easier than annotating. In this paper, we leverage unsupervised transfer learning to perform plant disease recognition, by which we achieve a better performance than supervised transfer learning in many cases. Besides, a vision transformer with a bigger model capacity than convolution is utilized to have a better-pretrained feature space. With the vision transformer-based unsupervised transfer learning, we achieve better results than current works in two datasets. Especially, we obtain 97.3% accuracy with only 30 training images for each class in the Plant Village dataset. We hope that our work can encourage the community to pay attention to vision transformer-based unsupervised transfer learning in the agricultural field when with few labeled images.

• The Journal of Korea Robotics Society
• /
• v.17 no.4
• /
• pp.425-430
• /
• 2022

As robots are considered one of the mainstream digital transformations, robots with machine vision becomes a main area of study providing the ability to check what robots watch and make decisions based on it. However, it is difficult to find a small object in the image mainly due to the flaw of the most of visual recognition networks. Because visual recognition networks are mostly convolution neural network which usually consider local features. So, we make a model considering not only local feature, but also global feature. In this paper, we propose a detection method of a small marker on the object using deep learning and an algorithm that considers global features by combining Transformer's self-attention technique with a convolutional neural network. We suggest a self-attention model with new definition of Query, Key and Value for model to learn global feature and simplified equation by getting rid of position vector and classification token which cause the model to be heavy and slow. Finally, we show that our model achieves higher mAP than state of the art model YOLOr.

• Journal of Korean Ophthalmic Optics Society
• /
• v.10 no.1
• /
• pp.71-82
• /
• 2005

The aim of this study was to provide fundamental data for the factors influencing the asthenopia of emmetropia with phoria and alleviation of asthenopia. A total of 348 subjects, aged between 19 and 30 years old, who had no strabismus, an eye trouble or whole body disease, were examined using corrected visual acuity, corrected diopter, stereopsis and suppression tests from September of 2002 to September of 2004. We excluded 21 subjects for the following reasons: if they had an amblyopia affecting binocular vision or inaccurate data. After these exclusions, 327 subjects remained. We then individually measured the refractive error correction, pupillary distance, optical center distance, phoria, convergence, accommodation and the AC/A as well as the asthenopia during binocular vision using a questionnaire. After analysis of factors affecting asthenopia, we also examined the reductive effect of a prism on the asthenopia in subjects who had asthenopia. To determine the factors affecting asthenopia during binocular vision, statistic analyses were carried out using the Chi-square test and the multivariate Logistic regression model. The results of this study were as follow. For asthenopia during near binocular vision of emmetropia with phoria, in case of the lower the accommodation and convergence, a significantly higher rate of asthenopia was observed (p<0.001). When the AC/A is lower, the higher the rate of asthenopia was observed but not significantly and there was no association between phoria and asthenopia. When the multivariate logistic regression model was used to determine factors affecting binocular vision of emmetropia with phoria, in case of the lower accommodation and convergence, a significantly higher rate of asthenopia was observed. when the phoria is esophoria or higher exophoria, or when the AC/A is lower than normal, the higher the rate of asthenopia was observed but not significantly and there was no association between phoria. AC/A and asthenopia. Therefore accommodation and convergence could be predictive factors for asthenopia during near distance binocular vision. Prism was used among' subjects who had asthenopia during near distance binocular vision, the symptom of asthenopia was eased up to 74.2% in emmetropia with phoria.

• The Journal of Korea Robotics Society
• /
• v.3 no.3
• /
• pp.194-202
• /
• 2008

This paper describes a new method for indoor environment mapping and localization with stereo camera. For environmental modeling, we directly use the depth and color information in image pixels as visual features. Furthermore, only the depth and color information at horizontal centerline in image is used, where optical axis passes through. The usefulness of this method is that we can easily build a measure between modeling and sensing data only on the horizontal centerline. That is because vertical working volume between model and sensing data can be changed according to robot motion. Therefore, we can build a map about indoor environment as compact and efficient representation. Also, based on such nodes and sensing data, we suggest a method for estimating mobile robot positioning with random sampling stochastic algorithm. With basic real experiments, we show that the proposed method can be an effective visual navigation algorithm.

• 제어로봇시스템학회:학술대회논문집
• /
• 1997.10a
• /
• pp.1840-1843
• /
• 1997

The task environment of a robot is changing rapidly and task itself becomes complicated due to current industrial trends of multi-product and small lot size production. A convenient user-interfaced off-line programming(OLP) system is being developed in order to overcome the difficulty in teaching a robot task. Using the OLP system, operators can easily teach robot tasks off-line and verify feasibility of the task through simulation of a robot prior to the on-line execution. However, some task errors are inevitable by kinematic differences between the robot model in OLP and the actual robot. Three calibration methods using image information are proposed to compensate the kinematic differences. These methods compose of a relative position vector method, three point compensation method, and base line compensation method. To compensate a kinematic differences the vision system with one monochrome camera is used in the calibration experiment.

• International Journal of Contents
• /
• v.14 no.1
• /
• pp.1-6
• /
• 2018

Text, as one of the most influential inventions of humanity, has played an important role in human life since ancient times. The rich and precise information embodied in text is very useful in a wide range of vision-based applications such as the text data extracted from images that can provide information for automatic annotation, indexing, language translation, and the assistance systems for impaired persons. Therefore, natural-scene text detection with active research topics regarding computer vision and document analysis is very important. Previous methods have poor performances due to numerous false-positive and true-negative regions. In this paper, a fully-convolutional-network (FCN)-based method that uses supervised architecture is used to localize textual regions. The model was trained directly using images wherein pixel values were used as inputs and binary ground truth was used as label. The method was evaluated using ICDAR-2013 dataset and proved to be comparable to other feature-based methods. It could expedite research on text detection using deep-learning based approach in the future.