In the mobile reception of OFDM, reception characteristics are degraded by ICI(inter-Carrier Interference). In ISDB-T, convolutional code and Reed-Solomon code as error-correcting codes are used. In this paper, decoding and encoding of the error-collecting code are operated in the iterative detection. As the results, it is possible to improve the characteristics.

抄録全体を表示

We have been developing a 4K video quality evaluation equipment which compares a test (encoded) video to a reference video in real time. Our system calculates an objective quality score (DSCQS: Double Stimulus Continuous Quality Scale) frame by frame considering the characteristics of the human eye.

抄録全体を表示

We have developed multiplexing and encoding technologies of multimedia services for 8K Super Hi-Vision satellite broadcasting. We describe a “General Data Transport Method” over MMT which realizes frequent data updating by push over broadcasting and prototype receiver installed under the method.

抄録全体を表示

We have proposed the Media - Unifying Platform (MUP) that enable s vi e wers to receive content in an appropriate delivery medi um according to a device function and user situation . This paper describes the method to detect the same content delivered by differ ent m edia by compari ng content metadata . It is confirm ed that the method is effective to be applied to MUP .

抄録全体を表示

We have studied a synchronous play back control and viewing system with Android TV sets. The experimental system showed that TV viewers can watch various video images from various viewing angles.

抄録全体を表示

We propose a method of photographs selection and layout for a photo album.Good photographs are selected using the expression of a smile in the photograph and the composition of the photograph.Photographs in the album are arranged order of flow of time.

抄録全体を表示

We propose a color correction method for underwater images. We use an exposure bracketing imaging for capturing appropriate color information of underwater scenes where red-components are poor. Using the obtained multi-exposed images, we correct color of underwater images via a multi-illuminant estimation. Experimental results show the effectiveness of our method.

抄録全体を表示

We previously reported the image sensor communication system which transmits information from a liquid crystal display (LCD) to a camera-equipped mobile device. In this system, two-dimensional Orthogonal Frequency Division Multiplexing (OFDM) with Quadrature Phase Shift Keying (QPSK) is used to generate information embedded images shown on the display. In this paper, usage of pilot signals is investigated for the purpose of improving reliability of the system.

抄録全体を表示

We propose a method for estimating reflectance spectra from tristimulus values by minimizing the squared error between the original tristimulus values and their estimated ones. The proposed least squares problem has an analytic solution. Experimental results show that the estimated reflectance spectra reproduce the original tristimulus values well.

抄録全体を表示

Autonomous robots are becoming widespread and expanding their applications. For such robots, it's necessary to understand the shape of surrounding objects and the position of obstacles precisely by range sensors such as 3D Vision Sensor. When measuring distance by 3D Vision Sensor, the problem is the measurement error caused by secondary reflection light and transmitted light. We studied a method for suppressing measurement error by using additional projection patterns with thin bright bars.

抄録全体を表示

An 8K 120 Hz image capturing experiment using a 133-Mpixel 60 Hz CMOS image sensor is described. To acquire an 8K 120 Hz image, 2:1 inter-line scanning and interpolation technique with adaptive noise reduction were adapted. Experimental results show the motion blur was improved and the resolution in the motionless area maintained compared with that of the conventional 60 Hz scanning. The signal-to-noise ratio was improved 4.4 dB by the adaptive noise reduction technique.

抄録全体を表示

In this paper we develop a prototype of a spherical audiovisual sensor. While the surrounding field of view is observed by a spherical camera consisting of a pair of fisheye cameras, the 3D distance of the surrounding space is coarsely acquired by 20 ultrasonic sensors distributed on the surface of a regular icosahedron.

抄録全体を表示

In this study, we propose a spectral computed tomography (CT) with reduced data traffic. Energy spectra taken from selected pixels were combined with conventional CT image to allow material identification , while reducing data traffic. S pectral CT image obtained by first generation CT system is being presented to verify our concept.

抄録全体を表示

We implemented an IoT sensor for evaluating comfortability of real estate. The IoT sensor can measure temperature, humidity, illuminance, smell and acceleration, and record sound and video. From the experiment in which we set one sensor in northern and southern rooms respectively, we succeeded in visualizing that the room temperatures were very different even in the same building. From the outdoor experiment, we could measure how the vibration and noise were detected in each place.

抄録全体を表示

Hilbert transform is required for Optical SSB (OSSB) modulation. We propose OSSB modulation using overlap Hilbert transform on frequency domain. We confirmed continuous transformed waveform and improved sideband suppression in OSSB spectrum.

抄録全体を表示

In this p aper , PLC under the impulsive noise environment is discussed . In the compensation of impulsive noise , it is necessary to estimate the channel . By computer simulation s , this paper s how s the good performance s of the proposed scheme using scattered pilot symbols under impulsive noise environment .

抄録全体を表示

D ual - polarized MIMO ultra - multilevel OFDM using BCH code and LDPC code are proposed for Super Hi - Vision(SHV) . In the mobile reception of OFDM signal, reception characteristics decline by inter - carrier interference(ICI). In this paper, the reception characteristics of dual - polarized MIMO ultra - multilevel OFDM using error correcting code under mobile reception are evaluate d.

抄録全体を表示

The compensation scheme for the interference that is caused by the delayed wave over Guard Interval is proposed. The proposed scheme is contracted by removing ISI component and extending artificial Guard Interval. In the results of computer simulations, it is possible to improve reception characteristics by the proposed scheme.

抄録全体を表示

We have studied the underwater optical video transmission system for live broadcasting from the submarine to the ship on the sea via the optical repeater. This report describes a met h od to know the position of the submarine from the ship with a laser rangefinder on the optical receiver module.

抄録全体を表示

We propose an optical transmission system for realizing TV live broadcasting from a submarine to a ship via a repeater that has receiver modules on its spherical surface. We made a small and stable optical module that can track incident light with wide range , and can send the submarine voice message，too.

抄録全体を表示

A novel sampling strategy suitable for compressed sensing MRI (CS-MRI) is proposed. Spatial resolution of reconstructed images is sometimes reduced due to variable density random undersampling. To improve spatial resolution in CSMRI, oblique scanning in k-space for the read-out direction is adopted to acquire a higher spatial frequency for the phaseencoding direction. Reconstruction experiments showed that spatial resolution can be improved without lengthening scan time.

抄録全体を表示

This paper propo ses a method to decide the appropriate region including shade for estimating illumination light from one image using log - chromaticity differences . Method of image segmentation adopts Ridge based Distribution Analysis which is not affected by distribution of shade.

抄録全体を表示

This paper proposes a measurement method of changes of body temperature by tracking the person in thermal images. We try to estimate more accurate changes of body temperature by considering the influence of room temperature , since body temperature is influence d by room temperature .

抄録全体を表示

Digitized picture doesn't have texture information such as the luster. Therefore its visual appearance can't be changed. This study proposes a method of viewpoint estimation by video image and acceleration sensor for representing texture of big picture, and tries to represent texture.

抄録全体を表示

A cquisition of high S/N image is essential for im plementation of high - performance 2D bar - code reader . E xper imental results show that low S/N image s cause the readable rate to decrease even if out - of - focus image s are deblurred . An exposure control technique fo r acquiring high dynamic rage images is also presented .

抄録全体を表示

HEVC Intra Coding requires considerable calculation. In this paper, we propose a method that accelerates Intra coding processing by using deep learning. We have attempted to make the network models for each size of PU.

抄録全体を表示

The CT images by P hoton - counting type X - ray CT ha ve been studied on the image quality. Compared to current integration mode imager s , this type of detector can yield clearer CT image under low - dose. In this report, CT image s by Photon - counting typ e detector are evaluated in association with the fluctuation of energy spectra .

抄録全体を表示

Fluorescent sensing has become widespread in healthcare or biochemistry. However , for sensing analytes labeled with several fluorochromes, conventional methods require to repeat sensing them by switchi ng optical filters. In this report, we propose a method that can be used for simultaneously sensing fluorochromes with multi - bandpass filters and image processing.

抄録全体を表示

In X - ray detector used for CT, material identification is not possible in widely used current mode image sensor. To do mate rial identification using current mode image sensor, we report a signal detection method depending on the energy of X - rays with the interaction of X - ray and substance.

抄録全体を表示

A flicker-free driving method at a 120-Hz frame frequency by using interlaced scanning and an electrical shutter is proposed. This method suppresses flicker in video, which is captured at a 120-Hz frame frequency and at a 100-Hz illumination-intensity variation frequency.

抄録全体を表示

We present a method to predict by lifestyle questionnaire the time to pregnancy. We can estimate the probability distribution of the time to pregnancy with high precision.

抄録全体を表示

In this paper, we compare four typical recommendation algorithms aiming at online lecture recommendation: collaborative filtering, matrix factorization, tf-idf, and word2vec. The experimental results using 1,958 videos in Schoo showed that the matrix factorization is the best (precision was 0.28@10), followed by collaborative filtering. On the other hand, content-based filtering did not work well for lecture recommendation.

抄録全体を表示

We propose a method of extracting TV program themes. Our approach detects burst words from TV programs and analyzes the time-series correlation between all pairs of them. In addition, we have developed a system for visualizing the appearance frequency of the keywords on the time axis for investigating the themes.

抄録全体を表示

We present visual information with auditory cues to provide enough pseudo-haptic feedback when a user grasps a virtual object with a bare hand. We examine effects of the auditory cues on perception of grasping a virtual object with a bare hand.

抄録全体を表示

Online dating is a popular way to search for a partner for marriage. Based on data from dating website, we propose an approach to evaluate people’s objective attractiveness. We predict people’s attractiveness by SVR, based on people’s profiles. The experimental results have shown high correlations between our predicted attractiveness scores and their real values.

抄録全体を表示

Attributes in various computer vision application are used as inter mediate representation. Especially in object recognition, attributes are defined as components of object, e.g. “ leg ” and “ furry ” in “ dog ” , “ stripe ” and “ V - Neck ” in “ T - Shirt ” . Thus localized information in image is effective to recognize attributes themselves . In this paper, we evaluated the influence of using localized information in training attributes.

抄録全体を表示

In the past, the diffuse reflection component has often been assumed to be Lambertian in many studies of photometric and multi-view stereo, which aim to extract 3D information from images. We propose a method to correct intensities of non-Lambertian images for directly using the existing stereo reconstruction methods. Our method obtains Lambertian images by estimating the incident angles relative to the surface normals from a single image.

抄録全体を表示

In this paper, we propose a method for simultaneously recovering the surface normals, the diffuse reflectance parameters associated with the simplified Oren-Nayar-Wolff model, and the illumination conditions from multiple images captured under different light dirrections.

抄録全体を表示

We propose a method to estimate a position of a light source by using a single RGB-D image. First, highlights on scene object surfaces are detected by using depth information and luminance histograms. Next, light vectors (incident angles) are computed from normal vectors of the detected highlights and viewing vectors. Finally, the intersection of light vectors is calculated for estimating a three-dimensional position of the light source.

抄録全体を表示

In this paper, we propose a novel object tracking method to avoid tracking failures. The proposed method is based on probabilistic state estimation, and uses Kernelized Correlation Filter (KCF) for evaluating a likelihood. The experimental result shows that the proposed method achieves the superior performance to state-of-the-art methods.

抄録全体を表示

In this paper, w e aim at developing imaging techni que for handing down expertise and skills in the manufacturing industry and propose the method that temporally establishes correspondences between two different videos including common scenes of manufacturing works by integrating object and motion features .

抄録全体を表示

Surrounding projection mapping has been attracting attention as a method of editing appearance of objects by using multiple projectors. It is difficult for us to find the placement of multiple projectors which achieve s high - quality projection for complex - shaped objects because t he positional relationship between the projector s and the object s greatly affects the quality of the projection . Therefore, i n this research , we propose a method to find an optimal placement of the multiple projectors for complex - shape d object s . In this method, a complex - shape d object is divided into multiple simpl e - shape d object s, and also the surface s of the object are classified based on their normal direction , for finding the effective candidates of projector’s direction.

抄録全体を表示