• Journal of Information Processing Systems
• /
• v.10 no.2
• /
• pp.283-299
• /
• 2014

Skin detection is used in many applications, such as face recognition, hand tracking, and human-computer interaction. There are many skin color detection algorithms that are used to extract human skin color regions that are based on the thresholding technique since it is simple and fast for computation. The efficiency of each color space depends on its robustness to the change in lighting and the ability to distinguish skin color pixels in images that have a complex background. For more accurate skin detection, we are proposing a new threshold based on RGB and YUV color spaces. The proposed approach starts by converting the RGB color space to the YUV color model. Then it separates the Y channel, which represents the intensity of the color model from the U and V channels to eliminate the effects of luminance. After that the threshold values are selected based on the testing of the boundary of skin colors with the help of the color histogram. Finally, the threshold was applied to the input image to extract skin parts. The detected skin regions were quantitatively compared to the actual skin parts in the input images to measure the accuracy and to compare the results of our threshold to the results of other's thresholds to prove the efficiency of our approach. The results of the experiment show that the proposed threshold is more robust in terms of dealing with the complex background and light conditions than others.

• Journal of Internet Computing and Services
• /
• v.4 no.4
• /
• pp.9-14
• /
• 2003

A GUI (graphical user interface) has been a dominant platform for HCI (human computer interaction). A GUI - based interaction has made computers simpler and easier to use. The GUI - based interaction, however, does not easily support the range of interaction necessary to meet users' needs that are natural. intuitive, and adaptive. In this paper, the modified BMA (block matching algorithm) is proposed to track a hand in a sequence of an image and to recognize it in each video frame in order to replace a mouse with a pointing device for a virtual reality. The HCI system with 30 frames per second is realized in this paper. The modified BMA is proposed to estimate a position of the hand and segmentation with an orientation of motion and a color distribution of the hand region for real - time processing. The experimental result shows that the modified BMA with the YCbCr (luminance Y, component blue, component red) color coordinate guarantees the real - time processing and the recognition rate. The hand tracking by the modified BMA can be applied to a virtual reclity or a game or an HCI system for the disable.

• 한국HCI학회:학술대회논문집
• /
• 2008.02b
• /
• pp.60-63
• /
• 2008

The sony Eytoy is developed on Playstation 2 using webcam for detecting human. A user see his appearance in television and become real gamer in the game. It is very different interface compared with ordinary video game which uses joystick. Although Eyetoy already was made for commercial products but the interface method still is interesting and can be added with many techniques like gesture recognition. In this paper, we have developed game interface with image processing for human hand and face detection and with game graphic module. And we realize one example game for busting balloons and demonstrated the game interface abilities. We will open this project for other developers and will be developed very much.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2016.07a
• /
• pp.51-54
• /
• 2016

In computer vision especially in image processing, it has become popular to apply deep convolutional networks for supervised learning. Convolutional networks have shown a state of the art results in classification, object recognition, detection as well as semantic segmentation. However, supervised learning has two major disadvantages. One is it requires huge amount of labeled data to get high accuracy, the second one is to train so much data takes quite a bit long time. On the other hand, unsupervised learning can handle these problems more cheaper way. In this paper we show efficient way to learn features for classification in an unsupervised way. The network trained layer-wise, used backpropagation and our network learns features from unlabeled data. Our approach shows better results on Caltech-256 and STL-10 dataset.

• Korean Journal of Remote Sensing
• /
• v.21 no.3
• /
• pp.229-242
• /
• 2005

Photogrammetry is a current method of GIS data acquisition. However, as a matter of fact, a large manpower and expenditure for making detailed 3D spatial information is required especially in urban areas where various buildings exist. There are no photogrammetric systems which can automate a process of spatial information acquisition completely. On the other hand, LiDAR has high potential of automating 3D spatial data acquisition because it can directly measure 3D coordinates of objects, but it is rather difficult to recognize the object with only LiDAR data, for its low resolution at this moment. With this background, we believe that it is very advantageous to integrate LiDAR data and stereo CCD images for more efficient and automated acquisition of the 3D spatial data with higher resolution. In this research, the automatic urban object recognition methodology was proposed by integrating ultra highresolution stereo images and LiDAR data. Moreover, a method to enable more reliable and detailed stereo matching method for CCD images was examined by using LiDAR data as an initial 3D data to determine the search range and to detect possibility of occlusions. Finally, intellectual DSMs, which were identified urban features with high resolution, were generated with high speed processing.

• Journal of Korean Society for Geospatial Information Science
• /
• v.13 no.3 s.33
• /
• pp.59-67
• /
• 2005

Building information is primary source in many applications such as mapping, telecommunication, car navigation and virtual city modeling. While aerial CCD images which are captured by passive sensor(digital camera) provide horizontal positioning in high accuracy, it is far difficult to process them in automatic fashion due to their inherent properties such as perspective projection and occlusion. On the other hand, LIDAR system offers 3D information about each surface rapidly and accurately in the form of irregularly distributed point clouds. Contrary to the optical images, it is much difficult to obtain semantic information such as building boundary and object segmentation. Photogrammetry and LIDAR have their own major advantages and drawbacks for reconstructing earth surfaces. The purpose of this investigation is to automatically obtain spatial information of 3D buildings by fusing LIDAR data with aerial CCD image. The experimental results show that most of the complex buildings are efficiently extracted by the proposed method and signalize that fusing LIDAR data and aerial CCD image improves feasibility of the automatic detection and extraction of buildings in automatic fashion.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.11
• /
• pp.105-114
• /
• 2015

This paper proposes methods to remove background area and segment region of interest (ROI) in chest X-ray images. Conventional algorithms to improve detail or contrast of images normally utilize brightness and frequency information. If we apply such algorithms to the entire images, we cannot obtain reliable visual quality due to unnecessary information such as background area. So, we propose two effective algorithms to remove background and segment ROI from the input X-ray images. First, the background removal algorithm analyzes the histogram distribution of the input X-ray image. Next, the initial background is estimated by a proper thresholding on histogram domain, and it is removed. Finally, the body contour or background area is refined by using a popular guided filter. On the other hand, the ROI, i.e., lung segmentation algorithm first determines an initial bounding box using the lung's inherent location information. Next, the main intensity value of the lung is computed by vertical cumulative sum within the initial bounding box. Then, probable outliers are removed by using a specific labeling and the pre-determined background information. Finally, a bounding box including lung is obtained. Simulation results show that the proposed background removal and ROI segmentation algorithms outperform the previous works.

• Korean Journal of Remote Sensing
• /
• v.39 no.6_3
• /
• pp.1693-1705
• /
• 2023

Rural areas, which account for about 90% of the country's land area, are increasing in importance and value as a space that performs various public functions. However, facilities that adversely affect residents' lives, such as livestock facilities, factories, and solar panels, are being built indiscriminately near residential areas, damaging the rural environment and landscape and lowering the quality of residents' lives. In order to prevent disorderly development in rural areas and manage rural space in a planned manner, detection and monitoring of hazardous facilities in rural areas is necessary. Data can be acquired through satellite imagery, which can be acquired periodically and provide information on the entire region. Effective detection is possible by utilizing image-based deep learning techniques using convolutional neural networks. Therefore, U-Net model, which shows high performance in semantic segmentation, was used to classify potentially hazardous facilities in rural areas. In this study, KOMPSAT ortho-mosaic optical imagery provided by the Korea Aerospace Research Institute in 2020 with a spatial resolution of 0.7 meters was used, and AI training data for livestock facilities, factories, and solar panels were produced by hand for training and inference. After training with U-Net, pixel accuracy of 0.9739 and mean Intersection over Union (mIoU) of 0.7025 were achieved. The results of this study can be used for monitoring hazardous facilities in rural areas and are expected to be used as basis for rural planning.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.3
• /
• pp.446-452
• /
• 2021

As a higher standard for food consumption is required, the consumption of chicken meat that can satisfy the subdivided food preferences is increasing. In March 2003, the quality criteria for chicken carcasses notified by the Livestock Quality Assessment Service suggested quality grades according to fecal contamination and the size and weight of blood and bruises. On the other hand, it is too difficult for human inspection to qualify mass products, which is key to maintaining consistency for grading thousands of chicken carcasses. This paper proposed the computer vision algorithm as a non-destructive inspection, which can identify chicken carcass parts according to the detailed standards. To inspect the chicken carcasses conveyed at high speed, the image calibration was involved in providing robustness to the side effect of external lighting interference. The separation between chicken and background was achieved by a series of image processing, such as binarization based on Expectation Maximization, Erosion, and Labeling. In terms of shape analysis of chicken carcasses, the features are presented to reveal geometric information. After applying the algorithm to 78 chicken carcass samples, the algorithm was effective in segmenting chicken carcass against a background and analyzing its geometric features.

• Journal of Korea Multimedia Society
• /
• v.22 no.5
• /
• pp.535-546
• /
• 2019

Recently, there have been a lot of researches on the whole face replacement system, but it is not easy to obtain stable results due to various attitudes, angles and facial diversity. To produce a natural synthesis result when replacing the face shown in the video image, technologies such as face area detection, feature extraction, face alignment, face area segmentation, 3D attitude adjustment and facial transposition should all operate at a precise level. And each technology must be able to be interdependently combined. The results of our analysis show that the difficulty of implementing the technology and contribution to the system in facial replacement technology has increased in facial feature point extraction and facial alignment technology. On the other hand, the difficulty of the facial transposition technique and the three-dimensional posture adjustment technique were low, but showed the need for development. In this paper, we propose four facial replacement models such as 2-D Faceswap, OpenPose, Deekfake, and Cycle GAN, which are suitable for the Cox platform. These models have the following features; i.e. these models include a suitable model for front face pose image conversion, face pose image with active body movement, and face movement with right and left side by 15 degrees, Generative Adversarial Network.