Recently, with the rapid development of Internet technology, Social Network Service (SNS) has blossomed into a worldwide popularity. In this paper, we present a novel application for friend recommendation on SNS websites. Unlike the questionnaire based friend recommendation scheme used nowadays (e.g. online dating sites, online matchmaking sites), we focus on the fact that most of the online users may be interested with the strangers whose appearances are somehow attractive according to their own preferences. Therefore, we present a friend recommendation system based on the appearances on photos, which aims to "make friend by the first sight". The system is built upon 5000 portraits photos as source dataset with another 50 photos as training set. Once the user provides rating to several photos in the training set, we first build his/her appearance preference model based on face detection and multi-features co-operation. Then, the images in the source are ranked according to different features respectively. Finally, the results of multi-features are fused via the method of Borda count. The system is a useful complement to the conventional psychological tests based friend recommendation scheme. It is easy to play with and of a lot of fun.

View full abstract

We have studied on a method for personal adaptation in TV content recommendation. We investigated a type of information which viewers recalled during or after watching TV, in order to know information giving cause for concern of the TV viewers. In this report, the investigation will be outlined.

View full abstract

In recent years, studies of location identification by mobile phone have been attracting a lot of attention. In this paper, we measure and examine the performance of mobile phone GPS using various types of terminals of different venders under some environments and study the performance improvement of GPS errors using multiple mobile phones.

View full abstract

We proposed a method for estimating of museum visitors' routes. It uses structure of passages in a museum and the locations detected by photographs which are taken by visitors.

View full abstract

Analysis of front faces of cars by computer covers various applications such as caricatures of cars on web pages, retrieval of a specific car image from database, car design, and so on. This paper presents a method to analyze features of front faces of cars quantitatively by principal component analysis for shapes of main parts and their arrangement.

View full abstract

In this paper, we propose a novel correspondence search technique for volume data based on 3D Phase-Only Correlation (POC), and discuss a personal identfication method using volume data. The proposed approach provides high accuracy correspondence breaking the limit of over voxel resolution and exhibits high performance for personal identification.

View full abstract

A support system for pocket billiards exhibits guiding lines which are the directions to shoot the cue ball in order to sink target balls into a pocket. A function to estimate 2D direction of a cue stick has been developed in order to realize a function to inform a player of the error in cue direction.

View full abstract

We are developing a social service model "teleda" that aim to compose horizontal connection between TV viewers. In this paper, we mention about a medium scale verification experiment using "teleda" and characteristics of participants' activity in the experiment.

View full abstract

A pair of similar pictures look completely different depending on situations. In this paper, new phenomenon are proposed, where red-green stripes inverted frame by frame become visible by adding a small motion inducing signal. Through examining the mechanisms, functions of the eye system is made clear, such as, modified ◇ eye characteristics, which leads to the verification of the algorisms of signal processing of visual perception

View full abstract

In recent years, television systems with a wide field of view that offer an elevated sense of presence have been proposed. They have higher spatial resolutions than conventional systems. On the other hand, the effects of increasing the temporal resolution or frame rate of television systems have not yet been sufficiently investigated, with the exception of the effects of frame interpolation at the receiver. On the assumption that people will enjoy TV with wide field of view displays at home in the future, we conducted a quantitative analysis of the effects of increased frame rate by a subjective evaluation of moving picture quality. The results indicated that an increase in frame rate from 60 frames per second (fps) to 120 fps improved the evaluated moving picture quality by 0.46 rank, and an increase from 120 to 240 fps improved moving picture quality by 0.23 rank, with statistical significance.

View full abstract

We examined whether the asymmetry in perception of expanding/contracting motion arised on a multiple-task situation. The stimuli of local-expanding and local-contracting pattern were used in Experiment1, global-expanding and global-contracting pattern were used in Experiment2. The results suggest that the advantage of perceiving expanding motion on a dual-task situation.

View full abstract

In order to make it possible to presume the unpleasantness invoked by shaking images, a system which estimates the unpleasantness degree was developed. The system implements a new analyzing algorithm based on the relationship between physical characteristics of images and unpleasantness. As a result of psychological estimation experiments with 32 shaking images, it was confirmed that differences of estimated unpleasantness degree between the developed system and the average subject are all less than the standard deviation of the experimental data.

View full abstract

Border detection between two stimuli of different luminance was investigated using red, green, and blue stimuli for a wide range of reference luminance in surround luminance of 0.001 and 500 cd/m^2 with and without artificial pupil. Decrease of threshold contrast with artificial pupil is most eminent for the blue light indicating the effect of spherical aberration exists without artificial pupil. Physiological model of underlying mechanism which can explain the change of contrast curves with surround luminance is derived.

View full abstract

In this paper, we propose an effective image ranking algorithm using eye tracking data. First, a directed graph is constructed by the trajectories of eye movement. Second the rankings of the images are calculated by the PageRank algorithm. At last, we examine the effectiveness of our method through several eye tracking experiments.

View full abstract

We examined experiment on attention time by using eye mark recorder. As a result, we understood audio influences the stopping time according to the area.

View full abstract

In taking hand-written memorandum, the taking process plays often important roles for recalling related information in addition of notations in the memorandum. In this report, we focus on the process through taking hand-written memorandum. Here, we propose image processing methods for tracking the pen-position in handwritten memorandum by using sunglasses type camera.

View full abstract

The position tracking of the pupils and corneal reflections is important to precisely and remotely detect eye-gaze. However, detection is difficult under intense illumination and fast head movement. We improve the tracking performance by using a high-speed camera and improving the light source for tracking.

View full abstract

We have proposed the robust pupil detection method that the detection is supported by the face pose determined from the 3D pupil and nostril positions which one camera estimates. The method is upgraded from one camera to the stereo-calibrated cameras to deal with the problem of the glasses reflections.

View full abstract

The autistic infants do not tend to watch their mother's eyes. The disease can be diagnosed by examining the infant's gaze distributions when the movie of mother's face is presented. The proposed system immediately and accurately detects the mother's pupil positions in the movie provided by the color camera.

View full abstract

In the gaze detection system using stereo-calibrated two cameras, a user-calibration method is proposed, in which the user fixate only one target. The method uses the angle between the two cameras seen from the pupil. Accuracy is examined by increasing the number of the camera to four.

View full abstract

In this paper, we propose a new codebook construction method for product quantization based approximate nearest neighbor search. In the proposed method, dedicated codebooks are created for product quantization considering the variation of distribution of residual vectors in adopting multiple assignment.

View full abstract

A real time encoding video-on-demand system incorporating video watermarking is surveyed. Identifiable video is available for 20 concurrent accesses to the service. Almost invisible watermarks can survive against image processing.

View full abstract

In this paper, we propose a new method of motion estimation for high frame-rate images and simulate it using synthetic images and real high frame-rate images. The result of the simulation shows the effectiveness of the proposed method.

View full abstract

This paper proposes a new method for restoring MPEG-compressed images, and experimentally evaluates its restoration performance. The structure of the MPEG encoder/decoder enables us to construct its MPEG-restoration algorithm with the processing techniques employed by our previously proposed JPEG-restoration method.

View full abstract

We propose a method of combination of explicit and implicit direct vector derivation with Decoder-side ME. First of all, the best candidate is selected with explicit signaling. Then, we search around best candidate with Decoder-side ME. Experimental results show coding gain up to 0.8% BD rate reduction compared to a method based on explicit signaling.

View full abstract

This report proposes an adaptive basis switching technique of two-dimensional (2-D) non-separable (NS) GenLOT for picture coding. The conventional DCT and DWTs are separable and have disadvantage in representing diagonal edges and textures of a given picture. In order to overcome the problem, the authors proposed 2-D NS GenLOT as a previous work. In this work, the adaptive local basis control of overlap and direction is investigated and the significance is verified.

View full abstract

We studied a new intra coding method for reducing decoder complexity. We compared our approach with H.264 under deblocking filter on or off in objective evaluation method and subjective assessment and confirmed superiority in low-bit-rate.

View full abstract

VHDL source generation program for Dynamically Reconfigurable VLD Circuit is designed. Applying this program to MPEG-1 Audio Layer III (MP3) code Table, Synthesizable VHDL source is generated.

View full abstract

In luma-based chroma intra prediction in HEVC Test Model, chroma samples are predicted from down-sampled luma samples of the same block by linear model relationship. We propose applying horizontal filtering on luma samples to improve prediction performance by removing horizontal aliasing effect. Experimental results using HM-3 show up to 0.7% BD rate improvement for chroma signal.

View full abstract

This paper presents a CMOS time-of-flight range image sensor using charge draining only modulation. This structure allows us to achieve high-speed charge transfer and reduce the noise that is caused by a transfer gate.

View full abstract

In this study, a ray-tracing algorithm for direct rendering of implicit surfaces is presented. The ray-surface detection process in the rendering algorithm is accelerated using the technique of piecewise polynomial approximation and parallel computation on the GPU.

View full abstract

We propose a new method of the generation of an artificial aerial image from the cartograph. We show that this problem is formulated as the texture synthesis problem with label constrain. This problem can be solved by using gibbs-sampler. The experimental results show that the proposed model can generate the aerial image from the input label image naturally.

View full abstract

In this paper, we propose a key frame detection method for key shot of news. Instead of using features specific to certain news topics, our proposal is to extract important features which are common to news topics, and to generally detect the key frame which represent each news topic.

View full abstract

In this paper, we propose a dynamic camera calibration method for zoom-camera whose angle and focus are changed every moment according to camera operations such as pan, tilt, and zoom. Our proposed method utilizes time correlations and estimates camera parameters based on feature point matching between video frames. Experimental results show the effectiveness of the proposed method.

View full abstract

In this paper, we propose a depth estimation method using the belief propagation. The feature of the proposed method is utilizing propagation characteristics and energy distribution to reduce estimation error which is caused by expandable effects, occlusion and no texture area. In experiments, we confirmed effectiveness of the proposed method by 3D modeling of sumo contents and their display on 3DTV using Integral Photography.

View full abstract

We previously proposed a method of reconstructing Light-Field from multi-focus images by dimension reduction and a 2-D filtering. It enables us to achieve conversion between 4-D Light-Field and 3-D multi-focus imaging sequences without significant degradation. In this paper, based on the method, Light-Field compression via synthesized multi-focus images is discussed.

View full abstract

We developed the adapter for 3D glasses when we used to measure the exact brightness of the 3D television through 3D glasses.

View full abstract

This paper describes a method of depth adjustment for 3D images. We developed the disparity estimation method and the image shifting method by down-scaling the L/R images horizontally on the horizontal borders basis, and implemented its algorithm to FPGA circuit. In addition, we evaluated these methods using a subjective assessment, and confirmed the effect of depth adjustment using our methods.

View full abstract

The depth of the stereoscopic images are changed by shifting L/R images horizontally. However we cannot perceive the change of the depth accurately because of the perceptual factors like size consistency. Therefore, we propose "Depth-scale guide" for the stereoscopic images. In a subjective experiment we confirmed the effect of "Depth-scale guide".

View full abstract

We have developed 5-inch QVGA flexible OLED displays driven by InGaZnO4 TFTs. The TFT backplane was formed on a plastic film substrate at temperatures below 130°C by applying a coatable.

View full abstract

A flexible organic light emitting diode (OLED) display driven by organic thin film transistors (OTFTs) has been successfully fabricated, after the materials and structures of image pixels on the OTFT backplane were improved to reduce defective pixels. We confirmed the enhancement of image quality in terms of uniformity on the flexible display.

View full abstract

We developed an air-stable low-voltage CaMgO protective layer. The luminous efficiency of a Ne+Xe (20%) PDP with the CaMgO protective layer is 1.2 times higher than that of the conventional Ne+Xe (10%) PDP with an MgO protective layer. The driving voltages and luminances of these PDPs are almost the same.

View full abstract

We developed a light source unit for laser TV. It is unified a laser module, laser power supply, a control circuit, optical systems and the light source unit provides stable optical power to laser TV.

View full abstract

We have developed Laser Backlighting LCD TV that uses the laser for one of the light sources. The developed LCD TV uses two kinds of light sources of the red laser and cyan LED. The developed LCD TV realizes a wide color gamut.

View full abstract

We realized a digital signage system that collects congestion degree around digital signage terminals and controls contents on other terminals based on the congestion degree.

View full abstract

This paper presents our hypothesis concerning viewer's attention to Digital Signage for Public Transportation and a first evaluation of a demonstration experiment at an airport.

View full abstract

We experimented the subjectivity evaluation to clarify the difference about a picture of a digital photo frame and a printed image of a photo printer. As a result, the high-performance model of the digital photo frame got the evaluation that was higher than the photo printer about a picture.

View full abstract

We have been studying a new audio-visual system providing super realistic experience to viewers in a dome theater. While a rear projection is essential for the enhancement of audio-visual service, an ordinary screen for rear projection is not acoustically transparent. Therefore, an acoustical transparent screen for rear projection was studied.

View full abstract

The purpose of this study is to examine the robustness of image quality factors for environment illuminations using the parameter design in the field of quality engineering. In this paper, image quality factors were gamma, maximum luminance and minimum luminance, and environment illuminations were D65 illumination, darkroom and indirect sunlight.

View full abstract