We have developed a network camera which is equipped with multi-lens optical system. Wide-angle observation is achieved by locating prisms on the optical axes of some of the lenses. The optical system of the camera is more compact than general wide-angle lenses and can get less distorted pictures. This camera can detect the pictures in which moving objects are captured, and can record only them. We have confirmed the robustness of the moving objects detection by experiments.

View full abstract

We propose a multi-sensor tracking approach for robust object tracking in a distributed camera environment. In our approach, we employed Wi-Fi based location estimation using wave strength from multiple Wi-Fi access points. Then, object tracking is conducted using particle filters using the location estimation results as supplementary information.

View full abstract

In a badminton game in recent years, construction of the strategy based on image information is important. We have studied the image information system to support for badminton strategy analysis. In previous work, image extraction of a player's upper half of the body from the video for game analysis was proposed. In this work, image processing for detection of the position information based on the center of gravity of a player's video is proposed. An experiment shows the result of having applied the proposed method to the image of the international first class match.

View full abstract

The watermarking based on Karhunen Loeve transformation of color space is the embedding method for utilizing the data structure of an image. This decoding algorithm is an opened type. This has a vulnerable of an exchange attack of the watermark to an embedding image. In this paper, the discrimination of embedding images is proposed, when a watermark is replaced. By the watermark embedding, the data structure of an embedding image becomes two parallel plane. Such character makes it possible to utilize the Linking Number (LN) as an index with which an image is expressed globally. The proposed method performs discrimination by LN of an original embedding image. The LN of experimental result that applied various kinds of image processing performed to the embedding image is shown and also considered.

View full abstract

We before proposed the technique that used 3D shape and 2D texture as this method of driver posture estimation. In this paper, it distinguished dangerously for driving by using the result of the posture estimation. We were able to detect the dangerous scene with accuracy about 78[%] that the driver drove without looking ahead.

View full abstract

Conventional show-through cancellation algorithm has a problem that show-through in an image area is excessively corrected, resulting in large error in a corrected result. In order to solve this problem, iterative algorithm is proposed and experimentally shown to be effective.

View full abstract

We propose a function-oriented classification of` web images and show new layout based retrieval system using this categorization. We defined nine categories of images taking into account of their functions used in web pages, and classified web images by using Support Vector Machine (SVM) in tree structured way. By using this classification of web images based on their roles, we constructed new layout based web page retrieval system. There is a strong correlation web image category with web layout, so we realized this system with high accuracy and based on human vision.

View full abstract

This paper describes a perceptual experiment on naturalness of replacing a phoneme segment to other speaker's one in a speech synthesis and the results of word speech synthesis by the multi speaker database.

View full abstract

We are developing the system that broadcasts the multi-viewpoint HD image that uses twelve HD cameras at a live broadcasting of sports. A smooth multi-viewpoint HD image can be generated by matching the target point of multi cameras by the projection transformation. This paper describes the system that does the projection conversion by GPU at high speed.

View full abstract

We have proposed a method to generate free viewpoint images based on a ray space representation. The method divides real space into local regions, generates free viewpoint image for each region, and integrates these images. The method enables horizontal walk-through experience. In this paper, we propose a method to generate free viewpoint images with vertical displacement. It is realized by adjusting display position of local free viewpoint pictures in the integration process. Computer simulation shows that the proposed method can reproduce natural vertical disparities among objects.

View full abstract

We suggest a new flow-motion generation method. In this method, we use Homographys, vanishing points, the size of reference object. And we made the system to track the positions of moving objects from multi-view images, and combined that system with our new Flow-motion method, and confirmed the flow-motion of moving objects.

View full abstract

We previously proposed methods for generating various images directly from multi-focus imaging sequences based on frequency analysis using 3-D FFT. In this paper, high-speed reconstruction of free viewpoint images by dimension reduction and 2-D FFT is discussed. We show experimental results by using synthetic and real images.

View full abstract

A motion segmentation technique for time-varying mesh (TVM) is presented. In our previous works, it has been pointed out that the motion analysis not in the feature vector space but in the 3D space is required for more accurate and robust motion analysis. Therefore, in this paper, we apply the iterative closest point (ICP) algorithm to the neighboring frames of TVM and the registration error is used to represent the degree of motion. The experiments demonstrated high precision and recall rates (>92%) is achieved.

View full abstract

In this paper, we describe synthesizing arbitrary view point images with 16 cameras system by using IBR. For synthesizing more realistic images, we propose the method to improve the accuracy of depth estimation by error judgment and re-estimation.

View full abstract

As an application of computational photography with variable-focus lens devices, this paper introduces a concept of adaptive IP imaging. It enables scene-adaptive aquistion of multiview images for the use of image-based rendering. Based on the virtual camera model, we describe how to adapt the imaging system to the target scene.

View full abstract

Integral Photography (IP) generates three-dimensional (3-D) images by elemental images at the position where the object was. We propose a method to control depth direction of 3-D image by processing elemental images. As a result, we confirmed that the proposed method can generate a 3-D image at an arbitrary position.

View full abstract

The properties of reconstructed images from Fresnel CGHs were investigated using a computer simulation method. It was possible to reconstruct a grayscale image from the CGH when interference fringes were recorded in a file with a 6-bit integer or more.

View full abstract

In this paper, we propose that by applying Wireless transmission technique, holographic data of 3-D moving objects are transmitted in high quality, and a relatively good reconstruction of the holographic images is performed.

View full abstract

Wide-gamut video system based on multispectral and multiprimary imaging can reproduce higher saturated colors than conventional color imaging systems. In this paper, we evaluate color reproducibility using iridescent colors and compare the results between 6-band system and 3-band system.

View full abstract

We have developed the 6 primary color PTV. Because of the newly developed 6-color color wheel and our color management technique, the newly developed PTV ensures a wide color gamut.

View full abstract

We compared the image quality of organic light-emitting diode (OLED) displays to that of LCDs by means of quantitative measurements of image quality parameters and subjective evaluations. The results indicate that the black-level and contrast performances of OLED displays in wide environments are superior to those of the LCDs. For these properties, subjective image quality of OLED display is especially high for fireworks images.

View full abstract

The penetration of flat panel displays, or FPD, has been rapidly spread in recent years. This paper describes the outline of FPD measurements carried out at NHK, and the results on the brightness, the contrast and the color temperature.

View full abstract

We measured a color reproduction performance of the flat panel displays, or commercially available consumer devices. The green and cyan color reproduction areas are extended for the some PDPs and LCDs. Some displays show large color reproduction errors on such colors.

View full abstract

We measured a moving image characteristics of the flat panel displays, or commercially available consumer devices. Results of measurements for MPRT (Moving Picture Response Time) and moving picture resolution, showed that PDPs are better than most of LCDs and 120Hz frame rate LCDs have the moving image characteristics similar to PDPs.

View full abstract

Frame rate conversion is required to solve blur at liquid crystal display and useful for low picture rate video signal. we discuss the determination of interpolation mode which has different motion vector, we use block smoothness to evaluate interpolation modes for better motion compensation.

View full abstract

TIR Fresnel lens are being used for wide angle optical engines of rear projection displays. In this report, we analyzed back scatterings of the lens under the general bright environment. We also propose a novel designing method of the lens that enables us to visually consider several boundary conditions at the same time.

View full abstract

Projection-displays have advantage in production equipments of facility of making screen large. However, the screen has problems in viewing angle, luminance uniformity etc.. These problems are depended on the diffusing property of the screen. We have optimized of the diffusing property of the screen, which enables to realize high-quality projection-displays.

View full abstract

In order to improve the image quality of rear projection display, it is necessary to establish the evaluation method of screen resolution for the first step. In this paper, we propose a new simple method to measure MTF (Modulation Transfer Function) of the screen by using a newly designed special filter.

View full abstract

Diffuser is often used on displays for varied purpose. Back Scattering characteristics of diffuser shows how diffuser scatters outside light, and it is important parameter for display contrast. We measured a refractive index of diffuser and precisely evaluated the Back Scattering characteristics of the diffuser with refractive index matching.

View full abstract

Measurements with an optical topography system were made in changes in cerebral blood flow in the cerebral cortex of the frontal lobe when speech on Japanese and Italian news broadcasts were played back at speeds of 0.5 times slower than the normal speed of speech (1.0 times), and at speeds of 1.5 times higher.

View full abstract

In addition to the sampled motion in TV, many visual phenomena in psychology, such as motion-induced color, are analyzed by spatio-temporal signal processing for human eyes using communications theory (e.g., modulation and filtering), independently of the function of the human brain. In this paper, these are examined by existing experimental data.

View full abstract

Experiments with pinholes show that ocular chromatic aberration is a principal factor of the hue assimilation and is a subsidiary factor of the lightness assimilation. Also, simple experiments shows that image blur is a factor of both hue and lightness assimilation.

View full abstract

In this paper, spatio-velocity contrast sensitivity functions (SV-CSF) were measured by changing stimuli sizes and viewing distances, and a viewing-condition-dependent model of the SV-CSF is build based on these measured data. The viewing-condition-dependency of the SV-CSF is modeled as a concave function of relative viewing distances.

View full abstract

In recent years, many researchers are studying to realize the much more high fidelity and quality moving image. For this purpose, we analyze the pursuing characteristics of eye movement and the perception characteristics of modulation transfer function while present moving image, which was taken with a high-speed camera.

View full abstract

We are aiming at realizing the high resolution moving image, so we compared the difference of the characteristics of the smooth pursuit eye movement between LCD and CRT.

View full abstract

In this paper, we propose a method to create a noise model for specific camera to estimate image noise accurately. The noise of an image depends on the characteristics of camera used to take the image. The experimental results show the improvement of noise estimation.

View full abstract

IR imaging is used in a variety of applications such as automobile, security systems and night vision. Many kinds of IR sensors have been developed. A thermopile IR sensor is suitable for low-cost applications. However, the output signal generated by a thermopile is very small and it can be easily deteriorated by both internal and external noises. In this paper a low noise high-gain readout circuit for thermopile IR image sensor is proposed, and the noise of the readout circuit including thermopile resistance is analyzed.

View full abstract

When multispectral images gain only low S/N ratio, colorimetric accuracy as well as image quality decrease in the reproduced images. To improve the image fidelity, this paper proposes a spatial-spectral 3D filter which consists of spatial restoration and spectral color estimation. The effectiveness of the 3D filter is confirmed through simulations and experiments.

View full abstract

Spectroscopic devices which don't destroy image are liquid crystal devices applied to Ryot filter or Fourier spectroscopy but these devices have several problems which are response time, incident angle, resolution and temperature characteristics. In this paper we propose spectroscopic system which overcomes these problems and discuss application to video camera.

View full abstract

Stacked cell structure of three organic photoconductive films which are sensitive to only one of the primary color components has been fabricated. The output signals from the stacked cell have shown good spectral selectivity, demonstrating the color separation in the vertically stacked structure of organic photoconductive films.

View full abstract

We proposed a new X-ray imager for super high space resolution. This imager consists of CdTe diode and field emitter array. We verified the operation principle of proposed device about one pixel. Moreover, we also manufactured 12×12 matrix structure field emitter array to fabricate X-ray imager.

View full abstract

We proposed that material discriminated X-ray CT with conventional X-ray tube and energy discriminated X-ray imager. We could perform identification of material by attenuation coefficient (CT value), but could not obtain atomic number. We considered this reason is influence of scattering X-ray.

View full abstract

To achieve the improvement of the light sensitivity, we have developed the microlens array for ultrahigh speed, high sensitivity CCD. We optimized the shape of microlens array with a simulator for this CCD and fabricated the microlens with stamping. Using this microlens, we can elevate the light sensitivity of this CCD approximately twice that of original.

View full abstract