• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.8 no.3
• /
• pp.702-707
• /
• 2004

This article is concerned with container pose estimation using CCD a camera and a range sensor. In particular, the issues of characteristic point extraction and image noise reduction are described. The Euler-Lagrange equation for gaussian and random noise reduction is introduced. The alternating direction implicit(ADI) method for solving Euler-Lagrange equation based on partial differential equation(PDE) is applied. The vertex points as characteristic points of a container and a spreader are founded using k order curvature calculation algorithm since the golden and the bisection section algorithm can't solve the local minimum and maximum problems. The proposed algorithm in image preprocess is effective in image denoise. Furthermore, this proposed system using a camera and a range sensor is very low price since the previous system can be used without reconstruction.

• Journal of Broadcast Engineering
• /
• v.24 no.5
• /
• pp.765-774
• /
• 2019

In this paper, we propose a point cloud matching algorithm for multiple RGB-D cameras. In general, computer vision is concerned with the problem of precisely estimating camera position. Existing 3D model generation methods require a large number of cameras or expensive 3D cameras. In addition, the conventional method of obtaining the camera external parameters through the two-dimensional image has a large estimation error. In this paper, we propose a method to obtain coordinate transformation parameters with an error within a valid range by using depth image and function optimization method to generate omni-directional three-dimensional model using 8 low-cost RGB-D cameras.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.16 no.6
• /
• pp.323-330
• /
• 2021

Visual object tracking is a challenging area of study in the field of computer vision due to many difficult problems, including a fast variation of target shape, occlusion, and arbitrary ground truth object designation. In this paper, we focus on the reinforced feature of the dynamic search area to get better performance than conventional discriminative model prediction trackers on the condition when the accuracy deteriorates since low feature discrimination. We propose a reinforced input feature method shown like the spotlight effect on the dynamic search area of the target tracking. This method can be used to improve performances for deep learning based discriminative model prediction tracker, also various types of trackers which are used to infer the center of the target based on the visual object tracking. The proposed method shows the improved tracking performance than the baseline trackers, achieving a relative gain of 38% quantitative improvement from 0.433 to 0.601 F-score at the visual object tracking evaluation.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.16 no.5
• /
• pp.209-214
• /
• 2021

Panoptic segmentation, which is now widely used in computer vision such as medical image analysis, and autonomous driving, helps understanding an image with holistic view. It identifies each pixel by assigning a unique class ID, and an instance ID. Specifically, it can classify 'thing' from 'stuff', and provide pixel-wise results of semantic prediction and object detection. As a result, it can solve both semantic segmentation and instance segmentation tasks through a unified single model, producing two different contexts for two segmentation tasks. Semantic segmentation task focuses on how to obtain multi-scale features from large receptive field, without losing low-level features. On the other hand, instance segmentation task focuses on how to separate 'thing' from 'stuff' and how to produce the representation of detected objects. With the advances of both segmentation techniques, several panoptic segmentation models have been proposed. Many researchers try to solve discrepancy problems between results of two segmentation branches that can be caused on the boundary of the object. In this survey paper, we will introduce the concept of panoptic segmentation, categorize the existing method into two representative methods and explain how it is operated on two methods: top-down method and bottom-up method. Then, we will analyze the performance of various methods with experimental results.

• Archives of Craniofacial Surgery
• /
• v.22 no.2
• /
• pp.110-114
• /
• 2021

Retrobulbar hemorrhage is a disastrous condition that can lead to permanent blindness. As such, rapid diagnosis and treatment are critical. Here, we report a patient who presented with retrobulbar hemorrhage following an orbital floor fracture. Restoration of inferior orbital wall with porous polyethylene implant was underwent. Four days after the orbital floor reconstruction, the patient smoked a cigarette outdoors in -3℃ weather. Cold temperature and smoking caused an increase in his systemic blood pressure. The elevated blood pressure increased intraorbital pressure to the extent of causing central retinal artery occlusion and exacerbated oozing. During exploratory surgery, 3 mL of hematoma and diffuse oozing without arterial bleeding were observed. Prompt diagnosis and treatment prevented vision impairment. Few studies have reported on the risk factors for retrobulbar hemorrhage. This case showed that daily activities, such as exposure to cold weather or tobacco smoking, could be risk factors for retrobulbar hemorrhage.

• Journal of Industrial Convergence
• /
• v.2 no.2
• /
• pp.3-30
• /
• 2004

This study intends to closely examine an evaluation of effectiveness of the HRM associated with interactions among employees(I), purpose recognition(P), role feedbacks(R), and motivative compensation(M). The following is a summary of the results of this research: From the study of the actual situation of the human resource management by the Korean firms, it appeared that the form of existence of vision systems possessed the form of having been written as documents and co-shared. And, although, in the area of motivation endowment and feedbacks, the level of motivation endowment regarding the comprising member by those in charge of departments and the level of reflection of the company's personnel policies in the execution of work by the departments were ordinary, the level of presenting the opinion of the departments regarding the company's personnel policies was shown to be low. And, the decisive elements of wages and the compensation standard were in the order of performance basis and long service. Also, it was analyzed that the leadership style of the CEO is the most important element that influences human resource management policies. And, it was analyzed that in the evaluation of the human resource management area the roles regarding evaluation and compensation, especially, the mutual interactions of the comprising members, feedbacks, and improvement activities are inadequate. And the managerial implications are discussed.

• Journal of Korea Multimedia Society
• /
• v.22 no.9
• /
• pp.1000-1010
• /
• 2019

Image semantic segmentation and dehazing are key tasks in the computer vision. In recent years, researches in both tasks have achieved substantial improvements in performance with the development of Convolutional Neural Network (CNN). However, most of the previous works for semantic segmentation assume the images are captured in clear weather and show degraded performance under hazy images with low contrast and faded color. Meanwhile, dehazing aims to recover clear image given observed hazy image, which is an ill-posed problem and can be alleviated with additional information about the image. In this work, we propose a deep multi-task network for simultaneous semantic segmentation and dehazing. The proposed network takes single haze image as input and predicts dense semantic segmentation map and clear image. The visual information getting refined during the dehazing process can help the recognition task of semantic segmentation. On the other hand, semantic features obtained during the semantic segmentation process can provide cues for color priors for objects, which can help dehazing process. Experimental results demonstrate the effectiveness of the proposed multi-task approach, showing improved performance compared to the separate networks.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.19 no.4
• /
• pp.55-63
• /
• 2019

Proximity sensors and rearview cameras for automobile safety are common, but many accidents are still occurring. Using the All Around View and object recognition algorithm to show the front, back, left, right and bottom of the vehicle, the sensor detects the presence of a living body when the vehicle starts or parks, and displays the outside of the vehicle on the screen. In addition, the object recognition algorithm is used to visualize the object by expressing the position of the object. In this way, we propose a strong safety system that can prevent human accidents caused by the vehicle by sensing, screen, and expression.

• Electronics and Telecommunications Trends
• /
• v.36 no.1
• /
• pp.43-53
• /
• 2021

Super-realistic content technology has recently attracted attention as a core of the "new normal" that can overcome the spatial constraints caused by pandemics. It is moreover the core that allows users in remote locations to meet and engage in various social, cultural, and economic activities based on a network. Content technology is rapidly spreading beyond the existing entertainment area to various industries as an innovative tool that can be used to overcome space-time constraints and improve the productivity of industrial sites, because reality and virtual reality are now super-connected with ultra-low latency. However, existing services such as teleconferencing and tele-collaboration do not provide a level of realism that replaces face-to-face services, and various technical requirements have been proposed to overcome this. The trends in core technologies such as XR twins, hyper-realistic reproduction, sensory interaction, and emotional recognition technology, which are necessary for interactive realistic content that leads to feelings, from reproduction to experience and emotion, are explained. In this article, our aim is to present the future of realistic content that enables human care and can even overcome psychological difficulties such as the "Corona blues".

• Journal of Multimedia Information System
• /
• v.8 no.2
• /
• pp.101-110
• /
• 2021

In computer vision, single-image super resolution has been an area of research for a significant period. Traditional techniques involve interpolation-based methods such as Nearest-neighbor, Bilinear, and Bicubic for image restoration. Although implementations of convolutional neural networks have provided outstanding results in recent years, efficiency and single model multi-scalability have been its challenges. Furthermore, previous works haven't placed enough emphasis on real-number scalability. Interpolation-based techniques, however, have no limit in terms of scalability as they are able to upscale images to any desired size. In this paper, we propose a convolutional neural network possessing the advantages of the interpolation-based techniques, which is also efficient, deeming it suitable in practical implementations. It consists of convolutional layers applied on the low-resolution space, post-up-sampling along the end hidden layers, and additional layers on high-resolution space. Up-sampling is applied on a multiple channeled feature map via bicubic interpolation using a single model. Experiments on architectural structure, layer reduction, and real-number scale training are executed with results proving efficient amongst multi-scale learning (including scale multi-path-learning) based models.