We have proposed the contour extraction method using Active Contour Model with Structuring Elements. The results of contour extraction are influenced from the shape of structuring elements. We applied the method into the extractions of retina layers from OCT, and evaluated the effects of Element shape using Active Contour Model with Structuring Elements.

抄録全体を表示

In order to recognize hand shapes in arbitrary postures, we propose a shape descriptor based on HOG (Histograms of Oriented Gradients). Experiment results indicate that a recognition rate of 88% is achieved by using the proposed descriptor with a SVM (Support Vector Machine) classifier.

抄録全体を表示

Download time was measured when video file was downloaded simultaneously from watching PCs. And switching time from channel to channel was measured.

抄録全体を表示

We propose a fast encoding algorithm for QR Codes in Broadcast Markup Language. In the algorithm QR codes can be generated quickly in receivers by encoding individual information except fixed string such as server address. The fixed string is encoded in a broadcasting station in advance.

抄録全体を表示

It has been reported that MB size extension is useful for ultra-high resolution video coding because of its high spatial correlation. This paper shows that the extension is also useful for scalable coding of such resolution video.

抄録全体を表示

We propose a fast method of reduced image generation in H.264/AVC. Reduced images are processed by reference pixels for intra prediction. As a result, reduced images are generated by 60% decoding time.

抄録全体を表示

This paper proposes a lossless re-encoding scheme for still images already compressed by the lossy mode of JPEG 2000. In this scheme, quantized wavelet transform coefficients are extracted from the JPEG 2000 data, and then losslessly re-encoded based on block adaptive prediction and adaptive arithmetic coding both of which exploit intra and inter-subband correlations.

抄録全体を表示

In this paper, global motion compensated prediction is evaluated under extended macroblock size scheme that is well known as the effective coding function especially for ultra-high resolution images. Experimental results show the coding performance for 7680 x 4320 full resolution pictures, and the desirable condition to maximize the effect by the global motion compensation is addressed.

抄録全体を表示

In this paper, a method that generates metrics using the probablilty density of two signals immediately before and after in time is proposed. Results from the simulations of bit error rate (BER) found that the proposed method made an approximate 1dB improvement over conventional methods.

抄録全体を表示

We measured propagation and transmission data such delay profile, received signal level, C/N and BER by using digital terrestrial TV broadcasting under the SFN environment. On the basis of the data, we discussed delay time, C/N and BER performance in the SFN environment.

抄録全体を表示

An optical retransmission system for 12- and 21-GHz satellite broadcasting radio and 10-Gbps-class PON signals is proposed. Transmission properties of the system have been calculated.

抄録全体を表示

In this paper, we report the necessary electronic filter characteristics to realize Add-Drop system. When the system can be realized, we can drop unnecessary video signal and add new video signal in the same channel.

抄録全体を表示

We propose an assistant signal that enables Super Hi-Vision (SHV) camera focusing by a low resolution viewfinder. The signal is generated from high spatial frequency components of SHV video. The simulation results showed that the signal is enough effective for SHV camera focusing.

抄録全体を表示

We have developed a system for choreography simulation by motion synthesis of body section. The user puts many motions at the same time by arbitrary timing and generates new motions.

抄録全体を表示

We propose a method to generate new motion data with emotion. We extract the difference of feature between normal motion and motion with emotion as emotion component using 3DP-type Fourier description. New motion is generated by adding the component to another motion.

抄録全体を表示

This paper describes CG character making tools for TVML. Using them, an ordinary user can easily make or edit CG characters.

抄録全体を表示

The present study is to propose the automatic storyboarding to support the movie production. We examined what linguistic cues in the scenario are needed to storyboard. The results implicates that the composition of storyboarding is able to be determined based on the verb of scenario.

抄録全体を表示

This is the image works of composition o f the 3D image and video image. The realistic old memories of the stone are generated from the reconstructed image of hologram " a stone " and the video image.

抄録全体を表示

This paper describes how to generate a non-distorted B-spline surface for the boundary information of a Cut-Out shape. Our method generates points outside of the boundary based on CBD(Cross Boundary Derivative), and generates a surface using approximation method.

抄録全体を表示

In this paper, we propose a new framework to obtain precise 3D model, in which likelihood evaluation is introduced for foreground extraction in a 3D model-silhouette feedback scheme. Experimental results show that the proposed framework is less sensitive to threshold parameters for separation of foreground.

抄録全体を表示

This study proposes a new projection type 3-D display system. The feature of the system is the screen which is constructed by pixeled grating array. The grating array can generate high density directional images and enable us to observe 3-D images without special glasses.

抄録全体を表示

For a lenticular 3D display, the parallax number increases if a lenspitch is enlarged. However, a gap of view image occurs between lenses. Our coarse lenticular 3D display with a 3D diffusion screen can reduce the gap of the image.

抄録全体を表示

In this paper, we proposed a lenticular 3D display with barriers. It can provide a wider view than conventional lenticular 3D display and can be realized easily.

抄録全体を表示

Usual automobile does not have features to give velocity information around. This often causes vehicles to crash from the back. In this research, we propose two types of velocity lamp to put on vehicle's number plate, which are evaluated by using apofocus optical driving simulator to validate its effect.

抄録全体を表示

This paper assesses accuracy of depth perception by stereopsis combined with DFD by a CG simulator of the crane game. Precision of task under DFD conditions is improved for those who performed poorly under monolayer conditions.

抄録全体を表示

A novel, efficient method for mesuring the quality of holograms in the sense of radiating energy is introduced and discussed.

抄録全体を表示