We are researching binaural reproduction over line array loudspeakers as one of the alternative ways of 22.2 multichannel sound reproduction in the home. We developed a binaural reproduction processor and a pair of line array loudspeakers. The binaural reproduction processor transforms audio signals of 8K Super Hi-Vision (8K) programs to binaural signals in real time. The pair of line array loudspeakers, nine loudspeaker units are contained in each housing, are set immediately above and below the TV screen. The upper line array loudspeaker reproduces binaural signals of surround channels in the top layer of 22.2 multichannel sound and the lower one reproduces that of the middle layer’s.

View full abstract

Most of the latest supervised methods of person re-identification (PRID) cannot be used for the situations in which the sufficient amount of training data is not available. In this study, we propose a completely unsupervised method of PRID using partial correlation coefficient, and verify its effectiveness by some experiments.

View full abstract

Person authentication using periocular images has a great potential for recognizing pedestrians in surveillance applications. This paper proposes a periocular recognition algorithm using phase-based correspondence matching with texture enhancement. Texture enhancement of periocular images can significantly improve recognition performance of periocular recognition. Through experiments using a public database, we demonstrate the effectiveness of the proposed algorithm compared with conventional algorithms.

View full abstract

This paper presents a sound field expansion method for the interactive viewing of panoramic videos based on a spectral modification between multiple channels, and we developed a panoramic video player for smartphones with the stereophonic sound changing synchrony with the range of views.

View full abstract

NHK has been conducting research on advanced terrestrial transmission system combined dual-polarized MIMO and high-order-modulation OFDM, in order to transmit large volume contents such as 8K Super Hi-Vision. This paper reports the required field strength measured by field experiments in Hitoyoshi area, during 2013-2017.

View full abstract

Japan Broadcasting Corporation (NHK) has been researching terrestrial transmission for 8K Super Hi-Vision (8K SHV). In the actual SFN environment, we succeeded in carrying 8K SHV signals by using an advanced SFN technology and an 8K HEVC encoder/decoder.

View full abstract

Wideband transmission system using FEC with lower code rate and successive interference canceler technology can improve robustness against co-channel interference so that it would be expected to improve a frequency efficiency of utilization. This paper reports possibilities and problems in case of adopting wideband system to broadcast model in japan.

View full abstract

We have proposed the Media - Unifying Platform that enable s vi e wers to receive content in an appropriate delivery medi um according to d evice function s and viewers' situation . Th e paper describes the method to generate content link and the descriptive method . And we report the result of implementation experiment for verifying the method .

View full abstract

Bit selection method in 5:9 modulation code of holographic data storage was considered. By increasing the correlation between the code map of 5-bit and 9-bit, both error rate and iteration characteristics were improved when using low density parity check code error correction.

View full abstract

We designed an optical roll-off filter which cuts the high frequency components of the data page with suppression of the interpixel interference. We experimentally showed that this filter could increase the SNR of the reproduced data comparing with the use of conventional rectangle spatial low-pass filter.

View full abstract

A current-driven domain wall(DW) motion in magnetic nanowire has been utilized in MRAM. In order to reduce the driving current of DW, we estimated the DW velocity dependence of the applied magnetic field direction from a simulation. Our analyses indicate that in-plane field is effective for fast DW driving.

View full abstract

This paper takes up 2-component mixture spherically-symmetric Gaussian distribution model with signal’s degradation and sparsity, for the statistical modeling of DFT coefficients of video signals. To abstract the essence from these massive modelling results, clustering in the model space is indispensable. This paper discusses the centroid computation in the model space, requisite for the clustering.

View full abstract

This paper take up the estimation of probability distribution model parameter for the mean separation type 3D Short-Time DFT coefficients. This paper also shows the simulation results which estimation the model parameters for the noisy moving images. Simulation results shows that the model parameter estimation is possible to some extent.

View full abstract

In late years, anyone comes to be able to take photographs easily. However, it is difficult to take photographs with good composition like what professional photographers take. In this work, we propose a novel method for generating a set of images with optimized composition from an omnidirectional panoramic image.

View full abstract

We propose an extension of existing saliency map generation methods to deal with fisheye images. It makes use of an overset grid consisting of two latitude-longitude grids.

View full abstract

Broadcasters create diverse TV programs. We have considered a system to create encounters with valuable information by providing data related to the program redesigned in appropriate with user's daily behavio r using smartphone and IoT devices. In this study, we developed a prototype service for automobile users.

View full abstract

We surveyed actual usage of online video services on TV and smartphone by web-based questionnaire for a thousand people to understand needs of cast functionality. In the result, the cast functionality is not popular as it is now, whereas we concluded that there is higher potential needs especially in users of online video services on smartphone.

View full abstract

Our technology allows encoders on the distributor side to change transmission rates precisely according to the reception situation. This can reduce sudden variations in the reproduced image quality.

View full abstract

We are researching 64APSK coded modulation for the purpose of expanding the transmission capacity for satellite transmission. In the conventional 64APSK coded modulation design based on channel capacity, the influence by non-linear transmission path was not considered. In this report, we describe the design method considering non-linear distortion and equalization performance and show that better transmission performance than the conventional one can be achieved with the proposed method.

View full abstract

Seam carving is one of the techniques to resize the image while keeping the impression. In this paper, we propose a method of Seam Carving for complex background video image. We utilize a method of object detector “Binarized Normed Gradients for Objectness” in order to extract the object area in the videos.

View full abstract

Optical flow is estimated by imposing invariant conditions of brightness and its gradient. However, it is difficult to adjust the weights of these conditions due to the convenience of numerical calculation. In this paper, we consider the adjustment of the weights based on the estimation method that guarantees numerical stability.

View full abstract

Automatic crack detection from concrete surface image has been put to practical use. However, Increasing the resolution of the image requires high processing costs. In this paper , we propose a method of concrete crack detection for high-resolution image. We reduce processing costs by detect ion crack region using hessian matrix.

View full abstract

We propose a method for estimating reflectance spectra from tristimulus values observed under a number of illuminations. We formulate the estimation problem as a constrained Eucliean norm minimization, and show the analytic solution. Experimental results show that the proposed method improves the accuracy of the estimated reflectance spectra compared with the previous method.

View full abstract

In this research, we propose the color-correction method that transforms from images including near-infrared light to normal images without infrared light. Because colors of clothes remarkably change in images including infrared light, we classify the image into the area and other area, and correct colors in these areas separately.

View full abstract

In the next-generation DTTB, 4096QAM is used to improve the transmission capacity. In the detection, the high accuracy of channel estimation is required. In this paper, the reception characteristics using channel estimation scheme of amplitude and delay time are evaluated. As the result, it is possible to adapt the channel estimation scheme in the next-generation DTTB.

View full abstract

This report proposes a method of reducing the power consumption of dedicated emergency wake-up receiver for ISDB-T broadcasting signals. Both the misdetection and false alarm probabilities increase as decreasing the reception power, then the receiver tends to wrong results. The proposal reduces the power consumption by determining these probabilities.

View full abstract

In order to operate the hybrid cast in practice, we examined the mechanism to seamlessly switch between 2K and 4K.

View full abstract

This practice for 4K live transmission provides “Easy, High quality, and Cost performance” and 4K broadcasting in ISDB-T will result in affordable and pragmatic 4K in terrestrial braodcast

View full abstract

We have developed MPEG - 4 ALS hardware encoder which support s high - resolution audio streams. By implementing only encoding tools available in MPEG - 4 ALS Simple Profile and optimizing DSP (Digital Signal Processor) program s, we achieved MPEG - 4 ALS hardware encod er with real - time processing for high - resolution audio streams.

View full abstract

I t is important for runners to know the proper running form . However, inexperienced runners don’t know w here the key parts in runner’s form are . In this study, we prop ose a system to analyze the running for m automatically, focusing on the position of joints in the body by u sing image processing .

View full abstract

We developed the head pose estimation system for football players. Its purpose is to obtain useful metadata for program production in broadcasters. In this paper, prototyping and program production application of the system are introduced. From this prototyping, a series of production procedure on the system is confirmed.

View full abstract

Dominant region using V oronoi diagram is a method for analysis of soccer . In this research , we propose a method for expanding 2D - dominan t region into 3D - dominant region. Furthermore, we obtain success probability of pass by using 3D - d o minant region.

View full abstract

In recent ye ar s , texting while walking has become a problem. We have proposed a texting while walking recognition method using Convolutional Pose Machines (CPMs). CPMs is an algorithm to estimate the pose of people in an image . Conventionally, coordinates of parts are used as feature vector for SVM. In this paper, we will propose the texting while walking recognition method using joint angles info r mation .

View full abstract

In this report, depth of field processing of existing images is performed using depth images. We propose an image processing method that automatically generates depth images and performs processing based on the depth of field on software, so that images are easily generated without requiring special equipment.

View full abstract

Super resolution (SR) is a technique to create a high resolution image from low resolution images. Enhancer is, a conventional image quality improving method, which uses a straight forward one path signal processing. In contrast most of SR techniques need iterations to improve images. Super resolution image reconstruction (SRR) and learning based super resolution (LBSR) are typical SR methods and require the iterations to create high resolution images. Similar technologies have been discussed in astronomy. Researches have been trying to decrease the out of focus blur in astrophotography by blind deconvolution (BD). Although BD is not well known among broadcasting engineers, BD has been used in astrophotography to decrease the out of focus blur for more than 40 years. BD creates high resolution image by a completely different algorithm from SRR and LBSR. In this report BD merits against SRR and LBSR are discussed.

View full abstract

We propose a method for depth super-resolution using spatio-temporal correlation of RGB-D videos. The use of spatio-temporal correlations of RGB-D video allows us to suppress false artifacts caused by the color image in synthesizing super-resolved depth images. Our experiments demonstrate the effectiveness of our method.

View full abstract

We propose a novel method of high-quality view interpolation without disclosing entire 3D information like DIBR based on prediction using stereo matching. In this paper, by effectively embedding disparity information only on unpredictable regions in multi-view images as frequency components, our method maintains compatibility with applications using no 3D information.

View full abstract

P roposed a more accurate seam carving by combining graph cut and Laplacian fil ter with seam carving

View full abstract

The biggest objective and subjective loss of quality using the widely spread bicubic algorithm is at the supposed sharp edges of an image. This study focuses on developing an algorithm to restore the sharpness by using un - sharp masking influenced by the intensity of edges in each area. Testing this algorithm on 97 monochrome standard images results in up to 1dB improvement PSNR in nearly all cases, showing that the quality improved objectively. Subjective testing results favoured all the filtered i mage compared to the bicubic picture.

View full abstract

This article presents our proposed multiplexing scheme for IP packets in next-generation terrestrial broadcasting system, which is now under development in Japan. It also shows our designed modulator interface from broadcasting station to SFN of transmitters that supports the IP multiplexing scheme. The functionalities of multiplexing IP packets are confirmed with our implemented equipment.

View full abstract