• Journal of KIISE:Software and Applications
• /
• v.36 no.8
• /
• pp.653-663
• /
• 2009

In this paper, we propose a blind spots elimination system using three cameras. A wide-angle camera is attached on trunk for eliminating blind spots of a rear-view mirror and two cameras are attached on each side-view mirror for eliminating blind spots of vehicle's sides. In order to eliminate blind spots efficiently, we suggest a method to build a panoramic mosaic view with two side images and one wide-angle rear image. First, we obtain an undistorted image from a wide-angle camera of rear-view and calculate the focus-of-contraction (FOC) in undistorted images of rear-view while the car is moving straight forward. Second, we compute a homography among side-view images and an undistorted image of rear-view in flat road scenes. Next, we perform an image registration process after road and background region segmentation. Finally, we generate various views such as a cylinder panorama view, a top view and an information panoramic mosaic view.

• The Journal of the Korea Contents Association
• /
• v.3 no.4
• /
• pp.17-22
• /
• 2003

In this paper, we propose the new side-view face detection method in color images which contain faces over one. It uses color and the geometrical distance between nose and chin. We convert RGB to YCbCr color space. We extract candidate regions of face using skin color information from image. And then, the extracted regions are processed by morphological filter, and the processed regions are labeled. Also, we correct the gradient of inclined face image using projected character of nose. And we detect the inclined side-view faces that have right and left 45 tips by within via ordinate. And we get 92% detection rate in 100 test images.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.10B
• /
• pp.1093-1103
• /
• 2009

In this paper, we propose a new depth map based distributed multi-view video coding algorithm through an efficient side information generation. A distributed video coding scheme corrects errors between an original image and side information generated at a decoder by using channel coding techniques. Therefore, the more accurate side information is generated, the better performance of distributed video coding scheme is achieved. In the proposed algorithm, a distributed video coding scheme is applied to multi-view video coding based on depth map. Side information is also generated from images of adjacent views through 3D warping by using a depth map and is also combined with MCTI(motion compensated temporal interpolation) which uses images on a temporal axis, and 3D warping. Experimental results show that side information generated by using the proposed algorithm has 0.97dB better average PSNR compared with using MCTI and 3D warping separated. In addition, 8.01% of average bit-rate has been decreased while the same PSNR in R-D curves is kept.

• Journal of Multimedia Information System
• /
• v.8 no.2
• /
• pp.85-92
• /
• 2021

In a face, there is much information of person's identity. Because of this property, various tasks such as expression recognition, identity recognition and deepfake have been actively conducted. Most of them use the exact frontal view of the given face. However, various directions of the face can be observed rather than the exact frontal image in real situation. The profile (side view) lacks information when comparing with the frontal view image. Therefore, if we can generate the frontal face from other directions, we can obtain more information on the given face. In this paper, we propose a combined style model based the conditional generative adversarial network (cGAN) for generating the frontal face from multi-view images that consist of characteristics that not only includes the style around the face (hair and beard) but also detailed areas (eye, nose, and mouth).

• Journal of KIISE:Computing Practices and Letters
• /
• v.7 no.2
• /
• pp.132-140
• /
• 2001

This paper proposes an efficient walktlu-ough animation from two images of the same scene. To make animation easily and fast, Tour Into the Picture(TIP) enables walkthrough animation from single image but lacks the reality of its foreground object when the viewpoint moves from side to side, and view morphing uses only 2D transition between two images but restricts its camera path on the line between two views. By combining advantages of these two image-based techniques, this paper suggests a new virtual navigation technique which enable natural scene transformation when the viewpoint changes in the side-to-side direction as well as in the depth direction. In our method, view morphing is employed only in foreground objects , and background scene which is perceived carelessly is mapped into cube-like 3D model as in TIP, so as to save laborious 3D reconstruction costs and improve visual realism simultaneously. To do this, we newly define a camera transformation between two images from the relationship of the spidery mesh transformation and its corresponding 3D view change. The result animation shows that our method creates a realistic 3D virtual navigation using a simple interface.

• Journal of the Korean Association of Oral and Maxillofacial Surgeons
• /
• v.27 no.5
• /
• pp.423-427
• /
• 2001

Objective: This study was prepared to figure out a certain dimension and morphology of the condyle at the central, medial and lateral aspects on MR images of asymptomatic volunteers, which could be comparable with those of the TMD patients' condyle. Materials: Sixty TMJs from 30 asymptomatic volunteers(15 male, 15 female) who had no clinical symptoms and no disc displacement on sagital and coronal view of MRI were served as normal. Method: MR images were taken from the asymptomatic volunteers and the dimension of the anteroposterior length, mediolateral width, height, convexities were measured through the images on the sagittal and coronal sections of mandibular condyle. Then, these data were collected and analyzed. Result: The mean value of anteroposterior length was $8.00{\pm}1.21mm$ at central section and mediolateral length was $21.40{\pm}2.32mm$ on coronal view. The anterior condylar length at medial side was the shortest and the convexity of anterior slop at the lateral side was proved to be the flattest among 3 sections. There were little dimensional and morphological differences at sagittal sections, but the mediolateral width of condyle at coronal section was significantly different between male and female. Conclusion: In sagittal sections, the anterior condyle length was shortest at medial side and the convexity of anterior slop was flattest at lateral side, and there were little dimensional and morphologic differences between male and female. In coronal section, male's condyle was more wider and flatter than female's.

• The Journal of the Korea Contents Association
• /
• v.9 no.5
• /
• pp.13-23
• /
• 2009

One of the main purpose of multiple-view image processing technology is support realistic 3D image to device user by using multiple viewpoint display devices and compressed data restoration devices. This paper proposes a multiple view image synthesis scheme based on RAM disk which makes possible to browse 3D images generated by applying effective composing method to real time input stereo images. The proposed scheme first converts input images to binary image. We applies edge detection algorithm such as Sobel algorithm and Prewiit algorithm to find edges used to evaluate disparities from images of 4 multi-cameras. In addition, we make use of time interval between hardware trigger and software trigger to solve the synchronization problem which has stated ambiguously in related studies. We use a unique identifier on each snapshot of images for distributed environment. With respect of performance results, the proposed scheme takes 0.67 sec in each binary array. to transfer entire images which contains left and right side with disparity information for high quality 3D image browsing. We conclude that the proposed scheme is suitable for real time 3D applications.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.12 no.6
• /
• pp.199-206
• /
• 2012

Attempt to use as a marker of natural objects directly in the real world, but there is a way to use the accelerometer of the smartphone, to convert the Frontal-View virtual, because it asks only the pitch of the camera, from the side there is a drawback that can not be converted to images. The proposed algorithm, to obtain the rotation matrix of axis 3 pitch, roll, yaw, we set the reference point of the yaw of the target image. Then, to compensate for the rotation matrix to determine Myon'inji any floor, wall, the ceiling of the target image. Finally, to obtain the homography matrix for obtaining the Frontal-View to account for the difference between the gyro sensor coordinate system and image coordinate system, so we can get the Frontal-View from the captured images through the projection transformation was designed. Was tested to convert Frontal-View the picture was taken in an environment smartphone environment surrounding floor, walls and ceiling in order to evaluate the conversion program Frontal-View has been implemented, in this paper, design and The conversion algorithm implementation, it was confirmed that to convert a regular basis Frontal-View footage taken from multiple angles.

• Imaging Science in Dentistry
• /
• v.31 no.4
• /
• pp.209-214
• /
• 2001

Purpose: To investigate the ability of double TMJ view by multifunctional panorama to view the bony components and the space of the temporomandibular joint. Materials and Methods: Ten dry skulls fitted with resin shims over the articular surface of the condyle were used to reproduce the temporomandibular joint space. Fine metal wires were attached to the three portions of contours of the condylar head and the articular eminence. With 10 dry skulls and 20 cases having TMJ dysfunction, double TMJ views by multifunctional panorama (Planmeca 2002 Proline CC) and transcranial views were taken, analyzed from the anatomical view point, and compared statistically in view of the widths of the posterior joint space and the condylar head. Results: In double TMJ view, the supero-anterior part of the condyle represented the lateral 1/3, the most superior part represented center portion, and the posterior part medial l/3 of the condyle. In maximum mouth opening, no other structures were superimposed with the condyle in double TMJ view. In double TMJ view, petrous bone was moderately superimposed with the superior part of the condyle and the posterior increment of angle exposure made wider the images of the articular eminence and the condyle. The tendency of reduction in the posterior joint space appeared in the side of TMJ dysfunction compared with the normal side. The posterior joint spaces in double TMJ view were statistically wider (p<0.05) than those in transcranial view. The correlation coefficient was 0.5179 between the widths of the posterior joint spaces in two radiographic views. Conclusions: Double TMJ view can be substituted for transcranial view in evaluating the TMJ dysfunction.

• The KIPS Transactions:PartB
• /
• v.15B no.1
• /
• pp.9-16
• /
• 2008

In this paper, we present an algorithm for synthesizing intermediate view image from a stereoscopic pair of images. An image of multiview is need for people in order to easily recognize 3D image. However, if many cameras are use for that, not only does system get more complicated but also transmission rating cause a big trouble. Hence, stereo images are photograph and issue on the sending side and algorithm to generate several intermediate view image is able to be use on the receiving side. The proposed method is based on disparity space image. First of all, disparity space image that is depicted by the gap of pixel followed by disparity of stereo image is generated. Disparity map is made by utilizing disparity space image for searching for optimal disparity path then eventual intermediate view image is generated after occlusion region which does not match is processed. Experimental results illustrate the performance of the proposed technique and we obtained a high quality image of more than 30 dB PSNR.