• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.10
• /
• pp.190-196
• /
• 2014

This paper proposes a mobile-optimized pedestrian detection method using Cascade of HOG(Histograms of Oriented Gradients) for ADAS(Advanced Driver Assistance System) on smartphones. In order to use the limited resource of mobile platforms efficiently, the method is implemented by the OpenCL(Open Computing Language) library, and its processing time is reduced in the following two aspects. Firstly, the method sets a program build option specifically and adjusts work group sizes as variety of kernels in the host code. Secondly, it utilizes local memory and a LUT(Look-Up Table) in the kernel code to accelerate the program. For performance evaluation, the developed algorithm is compared with the mobile CPU-based OpenCV(Open Computer Vision) for Android function. The experimental results show that the processing speed is 25% faster than the OpenCV hogcascade.

• Korean Medical Education Review
• /
• v.18 no.3
• /
• pp.125-131
• /
• 2016

Recent changes in educational paradigms that emphasize the performance or outcomes of education are redefining how learning objectives are being described as 'learning outcomes' in various academic disciplines. Medical education is not an exception to this trend. However, it has come to our attention that the key concepts and appropriate descriptions of learning outcomes have not been well understood among educators and that this lack of understanding has hindered our efforts to implement the practice in the field. This study aims to provide a direction to establish and describe learning outcomes by examining previous studies that have focused on setting learning objectives as well as learning outcomes. Setting and describing learning outcomes starts from reflection on the approach of behavioral learning objectives, which overemphasizes learner's acquired knowledge, skills, and attitude in each classroom rather than actual performance. On the other hand, the learning outcome approach focuses on what the learner is able to do as a result of a learning experience. This approach is more learner-friendly and encourages students to lead and be responsible for their learning process. Learning outcomes can best be described when the relevance of actual contexts and the hierarchy of learning objectives are considered. In addition, they should be in the form of context, task, performance, and level, as well as be planned with proper assessment and feedback procedures. When these conditions are met, the learning outcome approach is beneficial to students as it presents a curriculum that is more open to learners. Despite these advantages of the learning outcome approach, there is a possible concern that setting the learning outcomes and describing them can restrict evaluation to lower cognitive skills if the concept of learning outcome is narrowly interpreted or is set too low. To avoid such narrow applications, it is important for educators to understand the comprehensiveness of the learning outcome setting and to consider long-term outcomes embedded in an organizational vision rather than only short-term behavioral outcomes.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.19 no.4
• /
• pp.470-477
• /
• 2009

Modular reconfigurable robots with physical docking capability easily adapt to a new environment and many studies are necessary for the modular robots. In this paper, we propose a vision-based fuzzy autonomous docking controller for the modular docking robots. A modular docking robot platform which performs real-time image processing is designed and color-based object recognition method is implemented on the embedded system. The docking robot can navigate to a subgoal near a target robot while avoiding obstacles. Both a fuzzy obstacle avoidance controller and a fuzzy navigation controller for subgoal tracking are designed. We propose an autonomous docking controller using the fuzzy obstacle avoidance and navigation controllers, absolute distance information and direction informations of robots from PSD sensors and a compass sensor. We verify the proposed docking control method by docking experiments of the developed modular robots in the various environments with different distances and directions between robots.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.22 no.1
• /
• pp.1-8
• /
• 2018

Recently, the convolutional neural network shows high performance in many computer vision tasks. However, convolutional neural networks require enormous amount of operation, so it is difficult to adopt them in the embedded environments. To solve this problem, many studies are performed on the ASIC or FPGA implementation, where an efficient representation method is required. The fixed-point representation is adequate for the ASIC or FPGA implementation but causes a performance degradation. This paper proposes a separate optimization of representations for the convolutional layers and the batch normalization layers. With the proposed method, the required bit width for the convolutional layers is reduced from 16 bits to 10 bits for the ResNet-50 neural network. Since the computation amount of the convolutional layers occupies the most of the entire computation, the bit width reduction in the convolutional layers enables the efficient implementation of the convolutional neural networks.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.45 no.6
• /
• pp.133-141
• /
• 2008

As interactive multimedia have come into wide use, user interfaces such as remote controllers or classical computer mice have several limitations that cause inconvenience. We propose a vision-based pointing device to resolve this problem. We analyzed the moving image from the camera which is embedded in the pointing device and estimate the movement of the device. The pose of the cursor can be determined from this result. To process in the real time, we used the low resolution of $288{\times}208$ pixel camera and comer points of the screen were tracked using local optical flow method. The distance from screen and device was calculated from the size of screen in the image. The proposed device has simple configurations, low cost, easy use, and intuitive handhold operation like traditional mice. Moreover it shows reliable performance even in the dark condition.

• Journal of the HCI Society of Korea
• /
• v.10 no.2
• /
• pp.45-55
• /
• 2015

In recent years, keypoint recognition and tracking technologies are considered as crucial task in many practical systems for markerless augmented reality. The keypoint recognition and technologies are widely studied in many research areas, including computer vision, robot navigation, human computer interaction, and etc. Moreover, due to the rapid growth of mobile market related to augmented reality applications, several effective keypoint-based matching and tracking methods have been introduced by considering mobile embedded systems. Therefore, in this paper, we extensively analyze the recent research trends on keypoint-based recognition and tracking with several core components: keypoint detection, description, matching, and tracking. Then, we also present one of our research related to mobile augmented reality, named mobile tour guide system, by real-time recognition and tracking of tour maps on mobile devices.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.62 no.2
• /
• pp.257-263
• /
• 2013

In case of image observation equipments, CCTV for short distance visual field is usually installed and operated mostly as the means of crime-prevention. However, the extensive demand for monitoring problems in case of the increase in intelligent crimes and disasters has led to the necessity of the development of long-distance observation equipments embedded with Night View functions. In case of the Night View equipments, the relevant market is set up to be focused mostly on Thermal Observation Device(hereinafter, TOD), but some shortcomings such as the limitation of image visibility and excessive maintenance cost, etc. have actually caused the necessity of new long distance Night View equipment. Moreover there might follow lots of difficulties in long-distance visualization in the event that irregular reflection is generated by minute particles in the atmosphere such as fog, smog, and dust, etc. These factors are motivate the work presented in this study. Our study is aimed at the realization of Pulsed Laser Illuminator and newly proposed Range-Gated image acquisition technology. And also the implementation of Tracker for continuous trace of the objects of interest from the obtained sequence images.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2005.10a
• /
• pp.384-389
• /
• 2005

The objective on this project is to develop a cooperative Field Robot (FR), by using a customize Open Control Platform (OCP) as design and development process. An OCP is a CORBA-based solution for networked control system, which facilitates the transitioning of control designs to embedded targets. In order to achieve the cooperation surveillance system, two FRs are distributed by navigation messages (GPS and sensor data) using CORBA event-channel communication, while graphical information from IR night vision camera is distributed using CORBA Asynchronous Method Invocation (AMI). The QoS features of AMI in the network are to provide the additional delivery method for distributing an IR camera Images will be evaluate in this experiment. In this paper also presents an empirical performance evaluation from the variable chunk sizes were compared with the number of clients and message latency, some of the measurement data's are summarized in the following paragraph. In the AMI buffers size measurement, when the chuck sizes were change, the message latency is significantly change according to it frame size. The smaller frame size between 256 bytes to 512 bytes is more efficient fur the message size below 2Mbytes, but it average performance in the large of message size a bigger frame size is more efficient. For the several destination, the same experiment using 512 bytes to 2 Mbytes frame with 2 to 5 destinations are presented. For the message size bigger than 2Mbytes, the AMI are still able to meet requirement far more than 5 clients simultaneously.

• Korean Journal of Remote Sensing
• /
• v.18 no.6
• /
• pp.359-367
• /
• 2002

In this paper, a new embedded wavelet packet image coder algorithm is proposed for an effective image coder using correlation between partitioned coefficients. This new algorithm presents parent-child relationship for reducing image reconstruction error using relations between individual frequency sub-bands. By parent-child relationship, every coefficient is partitioned and encoded for the zerotree data structure. It is shown that the proposed wavelet packet image coder algorithm achieves low bit rates and rate-distortion. It also demonstrates higher PSNR under the same bit rate and an improvement in image compression time. The perfect rate control is compared with the conventional method. These results show that the encoding and decoding processes of the proposed coder are simpler and more accurate than the conventional ones for texture images that include many mid and high-frequency elements such as aerial and satellite photograph images. The experimental results imply the possibility that the proposed method can be applied to real-time vision system, on-line image processing and image fusion which require smaller file size and better resolution.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.23 no.1
• /
• pp.55-62
• /
• 2019

Convolutional neural networks (CNNs) show high performance in the computer vision, but they require an enormous amount of operations, making them unsuitable for some resource- or energy-starving environments like the embedded environments. To overcome this problem, there have been much research on accelerators or pruning of CNNs. The previous pruning schemes have not considered the architecture of CNN accelerators, so the accelerators for the pruned CNNs have some inefficiency. This paper proposes a new pruning scheme, Conv-XP, which considers the architecture of CNN accelerators. In Conv-XP, the pruning is performed following the 'X' or '+' shape. The Conv-XP scheme induces a simple architecture of the CNN accelerators. The experimental results show that the Conv-XP scheme does not degrade the accuracy of CNNs, and that the accelerator area can be reduced by 12.8%.