• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.9
• /
• pp.1703-1710
• /
• 2017

This paper propose a 3D point clouds (depth) security technique for digital holographic display service. Image contents encryption is a method to provide only authorized right owners with the original image information by encrypting the entire image or a part of the image. The proposed method detected an edge from a depth and performed quad tree decomposition, and then performed encryption. And encrypts the most significant block among the divided blocks. The encryption effect was evaluated numerically and visually. The experimental results showed that encrypting only 0.43% of the entire data was enough to hide the constants of the original depth. By analyzing the encryption amount and the visual characteristics, we verified a relationship between the threshold for detecting an edge-map. As the threshold for detecting an edge increased, the encryption ratio decreased with respect to the encryption amount.

• The KIPS Transactions:PartB
• /
• v.14B no.3 s.113
• /
• pp.191-198
• /
• 2007

In this paper, we present a marker detection and recognition method from camera image for a disabled person to interact with a server system which can control appliance of surrounding environment. It converts the camera image to a binary image by using multi-threshold and extracts contours of objects in the binary image. After that, it approximates the contours to a list of line segments. It finds rectangular markers by using geometrical features which are extracted from the approximated line segments. It normalizes the shape of extracted markers into exact squares by using the warping technique. It extracts feature vectors from marker image by using principal component analysis and then recognizes the marker. The proposed marker recognition system is robust for light change by using multi-threshold. Also, it is robust for angular variation of camera by using warping technique and principal component analysis. Experimental results show that the proposed method achieves 100% recognition rate at maximum for 21 markers and execution speed of 12 frames/sec.

• Journal of the Korea Society of Computer and Information
• /
• v.11 no.4 s.42
• /
• pp.97-107
• /
• 2006

In this paper, we present a method for the detection and recognition of rectangular markers from a camera image. It converts the camera image to a binary image and extracts contours of objects in the binary image. After that. it approximates the contours to a list of line segments. It finds rectangular markers by using geometrical features which are extracted from the approximated line segments. It normalizes the shape of extracted markers into exact squares by using the warping technique. It extracts feature vectors from marker image by using principal component analysis. It then calculates the distance between feature vector of input marker image and those of standard markers. Finally, it recognizes the marker by using minimum distance method. Experimental results show that the Proposed method achieves 98% recognition rate at maximum for 50 markers and execution speed of 11.1 frames/sec for images which contains eleven markers.

• Journal of the Korea Convergence Society
• /
• v.12 no.11
• /
• pp.45-51
• /
• 2021

With the development of deep learning, semantic segmentation methods are being studied in various fields. There is a problem that segmenation accuracy drops in fields that require accuracy such as medical image analysis. In this paper, we improved PSPNet, which is a deep learning based segmentation method to minimized the loss of features during semantic segmentation. Conventional deep learning based segmentation methods result in lower resolution and loss of object features during feature extraction and compression. Due to these losses, the edge and the internal information of the object are lost, and there is a problem that the accuracy at the time of object segmentation is lowered. To solve these problems, we improved PSPNet, which is a semantic segmentation model. The multi-scale attention proposed to the conventional PSPNet was added to prevent feature loss of objects. The feature purification process was performed by applying the attention method to the conventional PPM module. By suppressing unnecessary feature information, eadg and texture information was improved. The proposed method trained on the Cityscapes dataset and use the segmentation index MIoU for quantitative evaluation. As a result of the experiment, the segmentation accuracy was improved by about 1.5% compared to the conventional PSPNet.

• Proceedings of the IEEK Conference
• /
• 2002.06c
• /
• pp.231-234
• /
• 2002

This paper proposes a new approach to extract contour of moving object from compressed video stream. We segment the area of moving object by using motion vector and extract the motion object block from it. And then we describe the connectivity direction of outline moving block, detect the edge related to connectivity direction in the block and finally obtain the contour by connecting the edges. This can divide the moving object only with motion vector and detect the exact contour on the basis of the edge automatically. Also, we can reduce spending time using motion block and remove the noise with directional edge. The experimental results demonstrate the accurate and effective qualify of the proposed method.

• The KIPS Transactions:PartB
• /
• v.9B no.1
• /
• pp.99-104
• /
• 2002

Existing methods of object tracking use template matching, re-detection of object boundaries or motion information. The template matching method requires very long computation time. The re-detection of object boundaries may produce false edges. The method using motion information shows poor tracking performance in moving camera. In this paper, a robust object tracking algorithm is proposed, using projected motion and histogram intersection. The initial object image is constructed by selecting the regions of interest after image segmentation. From the selected object, the approximate displacement of the object is computed by using 1-dimensional intensity projection in horizontal and vortical direction. Based on the estimated displacement, various template masks are constructed for possible orientations and scales of the object. The best template is selected by using the modified histogram intersection method. The robustness of the proposed tracking algorithm has been verified by experimental results.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.38 no.6
• /
• pp.635-644
• /
• 2020

Object recognition, detection and instance segmentation based on DL (Deep Learning) have being used in various practices, and mainly optical images are used as training data for DL models. The major objective of this paper is object segmentation and building detection by utilizing multimodal datasets as well as optical images for training Detectron2 model that is one of the improved R-CNN (Region-based Convolutional Neural Network). For the implementation, infrared aerial images, LiDAR data, and edges from the images, and Haralick features, that are representing statistical texture information, from LiDAR (Light Detection And Ranging) data were generated. The performance of the DL models depends on not only on the amount and characteristics of the training data, but also on the fusion method especially for the multimodal data. The results of segmenting objects and detecting buildings by applying hybrid fusion - which is a mixed method of early fusion and late fusion - results in a 32.65% improvement in building detection rate compared to training by optical image only. The experiments demonstrated complementary effect of the training multimodal data having unique characteristics and fusion strategy.

• Proceedings of the KAIS Fall Conference
• /
• 2009.12a
• /
• pp.171-176
• /
• 2009

본 논문은 사용할 수 있는 헤어스타일의 수가 제한되는 문제를 해결하기 위한 것으로, 상호대화식 매팅을 이용한 2D 가상 컬러 헤어스타일러에 관한 것이다. 사전에 준비된 그래픽 헤어스타일 외에도 원하는 헤어스타일을 보유한 2D 실사 영상으로부터 상호대화식 매팅 기술을 사용하여 헤어스타일을 분리 추출한 후, 영상 간 픽 앤 드롭(pick-and-drop) 방식으로 옮겨와 두상에 부착한 다음, 필요시 헤어스타일의 컬러도 자유롭게 변경할 수 있는 기능을 제공함으로써 저비용으로 활용 가능한 헤어스타일의 수를 증대시킬 수 있다. 이때 헤어스타일의 분리 추출은 사용자가 전경 객체의 개략적 윤곽을 그려줌에 따라 점증적으로 알파 매트를 계산하는 상호대화식 매팅 기술을 사용한다. 그리고 헤어스타일의 컬러 변경은 명도 차분 맵(intensity difference map)에 기반한 모발 컬러 매핑 기술을 사용한다. 제안된 방법은 직관적이고 편리한 상호대화식 사용자 인터페이스를 제공하기 때문에 작업자의 피로도를 경감시킴과 동시에 작업 시간을 단축할 수 있고 비숙련자도 간단한 사용자 입력을 통해 자연스러운 가상 헤어스타일을 생성할 수 있는 장점이다.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.10
• /
• pp.150-161
• /
• 2013

This paper presents a new system which produces high resolution 3D contents by capturing multiview images of an object using multiple cameras, and estimating geometric and texture information of the object from the captured images. Even though a variety of multiview image-based 3D reconstruction systems have been proposed, it was difficult to generate high resolution 3D contents because multiview image-based 3D reconstruction requires a large amount of memory and computation. In order to reduce computational complexity and memory size for 3D reconstruction, the proposed system predetermines the regions in input images where an object can exist to extract object boundaries fast. And for fast computation of a visual hull, the system represents silhouettes and 3D-2D projection/back-projection relations by chain codes and 1D homographies, respectively. The geometric data of the reconstructed object is compactly represented by a 3D segment-based data format which is called DoCube, and the 3D object is finally reconstructed after 3D mesh generation and texture mapping are performed. Experimental results show that the proposed system produces 3D object contents of $800{\times}800{\times}800$ resolution with a rate of 2.2 seconds per frame.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.11
• /
• pp.103-111
• /
• 2011

In this paper, we propose a Geometrical Centroid Contour Distance(GCCD) which is described by shape signature based on contour sequence. The proposed method uses geomertrical relation features instead of the absolute angle based features after it was normalized and aligned with dominant feature of the shape. Experimental result with MPEG-7 CE-Shape-1 Data Set reveals that our method has low time/spatial complexity and scale/rotation robustness than the other methods, showing that the precision of our method is more accurate than the conventional desctiptors. However, performance of the GCCD is limited with concave and complex shaped objects.