Dong-Su Lee

DISTANCE ESTIMATION OF CONCEALED OBJECTS WITH STEREOSCOPIC PASSIVE MILLIMETER-WAVE IMAGING

Millimeter waves can be used to detect concealed objects because they can penetrate clothing. Therefore, millimeter wave imaging draws increasing attention in security applications for the detection of objects under clothing. In such applications, it is critical to estimate the distances from objects concealed in open spaces. In this paper, we develop a segmentation-based stereo-matching method based on passive millimeter wave imaging to estimate the longitudinal distance from a concealed object. In this method, the concealed object area is segmented and extracted by a k-means algorithm with splitting initialization, which provides an iterative solution for unsupervised learning. The distance from a concealed object is estimated on the basis of discrepancy between corresponding centers of the segmented objects in the image pair. The conventional stereo-matching equation is modifled according to the scanning properties of the passive millimeter wave imaging system. We experimentally demonstrate that the proposed method can accurately estimate distances from concealed objects.

VECTOR CLUSTERING OF PASSIVE MILLIMETER WAVE IMAGES WITH LINEAR POLARIZATION FOR CONCEALED OBJECT DETECTION

Passive millimeter (MMW) imaging can penetrate cloth- ing to create interpretable imagery of concealed objects. However, the image quality is often restricted by low signal to noise ratio and temperature contrast as well as low spatial resolution. In this paper, we explore a four-channel passive MMW imaging system operating in the 8 and 3mm wavelength regimes with linear vertical and horizon- tal polarization directions. Both registration between difierent channel images and segmentation of concealed objects are addressed. Multi- channel image registration is performed by geometric feature matching and a-ne transform, and then multi-level segmentation separates the human body region from the background, and concealed objects from the body region, sequentially. In the experiments, several metallic and non-metallic objects concealed under clothing are captured in indoors. It will be shown that our method can separate objects with higher accuracy than the conventional method.

Image Registration and Fusion between Passive Millimeter Wave Images and Visual Images

Passive millimeter wave imaging has the capability of detecting concealed objects under clothing. Also, passive millimeter imaging can obtain interpretable images under low visibility conditions like rain, fog, smoke, and dust. However, the image quality is often degraded due to low spatial resolution, low signal level, and low temperature resolution. This paper addresses image registration and fusion between passive millimeter images and visual images. The goal of this study is to combine and visualize two different types of information together: human subjects identity and concealed objects. The image registration process is composed of body boundary detection and an affine transform maximizing cross-correlation coefficients of two edge images. The image fusion process comprises three stages: discrete wavelet transform for image decomposition, a fusion rule for merging the coefficients, and the inverse transform for image synthesis. In the experiments, various types of metallic and non-metallic objects such as a knife, gel or liquid type beauty aids and a phone are detected by passive millimeter wave imaging. The registration and fusion process can visualize the meaningful information from two different types of sensors.

Three-Dimensional Object Reconstruction and Recognition Using Computational Integral Imaging and Statistical Pattern Analysis

In this paper, we discuss computational reconstruction and statistical pattern classification using integral imaging. Three-dimensional object information is numerically reconstructed at arbitrary depth-levels by averaging the corresponding pixels. The longitudinal distance and object boundary are estimated where the standard deviation of the intensity is minimized. Fisher linear discriminant analysis combined with principal component analysis is adopted for the classification of out-of-plane rotated objects. The Fisher linear discriminant analysis maximizes the class-discrimination while the principal component analysis minimizes the error between the original and the restored images. The presented method provides promising results for the distortion-tolerant pattern classification.

Shape Feature Analysis of Concealed Objects with Passive Millimeter Wave Imaging

Millimeter wave (MMW) imaging has found rapid adoption in security applications such as concealed object detection under clothing. However, the imaging quality is often degraded due to resolution limit and low signal level. This study addresses shape feature analysis following concealed object detection. The object region is extracted by multi-level segmentation. Shape features are composed of several descriptors which are object area, perimeter, major and minor axes of the basic rectangle, rectangularity, compactness, and eccentricity. In the experiments, three objects (gun, hand ax, and plastic bottle containing liquid skin aid) concealed under clothing are captured by the passive MMW imaging system. The extracted shape features are compared with the true features from the object model showing good accuracy.

Polarized Light for Measuring a Human Skin Feature Indicating Aging

The textures of skin on the back of the hand of many men of different ages are analyzed to determine their changes with age. The analysis shows that the textures are segmented by many lines of different lengths and orientations, and the size and number of segments in the skin surface change with age; that is the size increases while the number decreases. As a result, the texture becomes less complex with age. However, the addition of wrinkles from the age of 50 onwards for certain groups of people makes the texture appear somewhat complex.

3D reconstruction method based on time-division multiplexing using multiple depth cameras

This article proposes a 3D reconstruction method using multiple depth cameras. Since the depth camera acquires the depth information from a single viewpoint, it’s inadequate for 3D reconstruction. In order to solve this problem, we used multiple depth cameras. For 3D scene reconstruction, the depth information is acquired from different viewpoints with multiple depth cameras. However, when using multiple depth cameras, it’s difficult to acquire accurate depth information because of interference among depth cameras. To solve this problem, in this research, we propose Time-division multiplexing method. The depth information was acquired from different cameras sequentially. After acquiring the depth images, we extracted features using Fast Point Feature Histogram (FPFH) descriptor. Then, we performed 3D registration with Sample Consensus Initial Alignment (SAC-IA). We reconstructed 3D human bodies with our system and measured body sizes for evaluating the accuracy of 3D reconstruction.

Comparative analysis of techniques to pick-up hologram data from real object

Holography is one method to record the information from a real scene, but it requires coherent illumination and the lack of resolution of the pick-up device may strongly limit the size of the object recorded. It is also possible to generate a hologram of a real scene in incoherent illumination condition by using techniques likes integral imaging or multiple imaging, but the spatial resolution provided from these methods is usually quite poor. Hologram can be made from a virtual scene with a computer, but the heavy computational load limit the size of the scene, and it is difficult to create precise models of complicated objects. In this paper, we analyze the different techniques used to pick-up 3D data from a real object such as holography or integral imaging. And then, we present the first result of a simulator which is developed to evaluate the key parameters of hologram data according to the pick-up system. Preliminary results may be possible to evaluate their performance and to choose the optimal method one should use according to the resolution, the depth of field or the angle of view.

Effect of Viewing Distance on 3D Fatigue Caused by Viewing Mobile 3D Content

With an advent of autostereoscopic display technique and increased needs for smart phones, there has been a significant growth in mobile TV markets. The rapid growth in technical, economical, and social aspects has encouraged 3D TV manufacturers to apply 3D rendering technology to mobile devices so that people have more opportunities to come into contact with many 3D content anytime and anywhere. Even if the mobile 3D technology leads to the current market growth, there is an important thing to consider for consistent development and growth in the display market. To put it briefly, human factors linked to mobile 3D viewing should be taken into consideration before developing mobile 3D technology. Many studies have investigated whether mobile 3D viewing causes undesirable biomedical effects such as motion sickness and visual fatigue, but few have examined main factors adversely affecting human health. Viewing distance is considered one of the main factors to establish optimized viewing environments from a viewers point of view. Thus, in an effort to determine human-friendly viewing environments, this study aims to investigate the effect of viewing distance on human visual system when exposing to mobile 3D environments. Recording and analyzing brainwaves before and after watching mobile 3D content, we explore how viewing distance affects viewing experience from physiological and psychological perspectives. Results obtained in this study are expected to provide viewing guidelines for viewers, help ensure viewers against undesirable 3D effects, and lead to make gradual progress towards a human-friendly mobile 3D viewing.

Advantage of diverging radial type for mobile stereo camera

Distortions in the perceived image characteristics for three different camera arrangements of parallel, converging, and diverging are different according to each focal length, focus distance, field of view angle, color, magnification, and camera aligning direction. The distortions in perceived image for the parallel and converging arrangements have been researched commercially available stereoscopic TV based on high speed LCD, shutter glasses, and mobile devices. However, the distortion in the perceived image for diverging arrangement is not well known. This paper discusses the distortion in perceived image characteristics of diverging type stereo camera according to the magnification determining the enlargement and reduction of a camera image, and they are compared with those of other camera arrangements such as parallel and converging types. Also, the distortion induces the image closer to the viewers for the diverging type while away for the converging. The inducement is more prominent as the camera distance between two component cameras of the stereo camera for the diverging type. Furthermore, the effect of diverging angle on disparity will be considered that the inter-camera distance can be made as small as possible.