PinP(Person is Presentation) is a new technique of rich content. We developed a PinP web content with motion capture system for promotion of Hokkaido Information University.

View full abstract

This paper describes a system which can synthesize a camera video and a CG scene corresponding to movement of camera using the newly developed TVML player.

View full abstract

We have proposed a novel method for free viewpoint generation in which "walk-through" view experience can be realized by using multiple local ray-spaces representation. In this paper, we extend our approach to utilize a zenith camera as well as horizontal camera array. Simulation result shows that we can successfully navigate the viewpoint vertically and horizontally in 3D space.

View full abstract

This paper presents a method of 3D reconstruction of human shape from multi-view videos. The 3D reconstruction method employs Graph Cut considering constraint by silhouette images. The result shows the method is effective against human object with complex costume

View full abstract

For reducing the contradiction between convergence and accommodation of the eyes, it is necessary to induce not only convergence but also accommodation properly In this paper it is verified with optometric experiment that induction of proper accommodation is realized using cylindrical lenses and high frequency striped patterns

View full abstract

Automobile drivers usually gaze at far distance therefore HUD for assistance of drive is the best to be apofocus optical system. In this research, we introduce prototype system to achieve assistance of drive system at night by integrating the picture from stereo camera and HUD using coarse integral imaging system.

View full abstract

The recognition of 3-dimensional coordinates is essential if robots are to truly help people in daily life. We propose a method that can hypothesize a 3-dimensional space from the relative movement speeds detected in video images Our approach is resolve human beings as moving object, and other things as static object. Then the robot will be able to follow human beings with constant distance

View full abstract

In tihis paper, we propose the robots system equipped with three robots, and examined the establishment of the extendibility of the group actions. The formational action in the robots arrangement that line up behind each other was examined by using this robots system.

View full abstract

This prompted us to conduct a trial to confirm the margins of error in the current GPS services available on mobile phones. It was carried out in several sites in Japan in order to ascertain the nature of the errors.

View full abstract

We propose the hand gesture interface to operate a wearable PC in outdoor using image processing. To recognize the hand movement, Hand area is expanded from the center point of hand. Next, the shape of hand is recognized by finger position and the center of hand. And, the shape is acquired in the time series.

View full abstract

Recently, various internet businesses are expanded. Especially, the business form that combines the store sales and the net sales attracts attention. It is said that such a business form obtains the synergy effect. This research aims to extract features of the store sales and the net sales.

View full abstract

In video delivery systems, it is important to protect to use video contents against the law according to DRM technologies. We propose a video delivery system with contents correctness verification function. The video data structure and authentication are focused on. After the system procedure is indicated, we consider the effect of utilization in this system.

View full abstract

We have already proposed MPF(Metadata Production Framework) where we can generate content-based metadata for video contents. One of our next targets is utilizing News video archives to provide new services based on the metadata. For this purpose, we proposed another concept where we can develop the News related services effectively.

View full abstract

The moral of TV stations has so collapsed that they often accept commercial messages of fraudulent businesses deceiving the consumers. This paper proposes a system which automatically removes such commercial messages. We propose how to cut the time of comparing registered CM images with TV stream.

View full abstract

We are developing a tool that can synthesize concatenate word speech and correct its degradation in order to make it broadcasting quality. In this study, several correction functions in were introduced into this tool. It is available to investigate better correction procedure to generate high quality synthesized speech.

View full abstract

This paper describes a method of constructing multi-agent models of Orchestral Musicians which play instruments under the interaction of musical information of a conductor and musicians. The experimental results using prototype software are also shown.

View full abstract

IA(Immune Algorithm), template matching, road signs

View full abstract

Signboards provide important information to guide and regulate the behaviors of pedestrians to make their course easily and more comfortable. Therefore, we propose a method for extracting a character sequence due to image features in color scene image.

View full abstract

The flow of freshwater into the sea is a key factor for understanding the hydrological cycle in both the sea and land regions. This study detects groundwater discharge points arising due to the difference in freshwater and seawater in coastal regions around Mt.Chokaisan by using remote sensing data.

View full abstract

In order to extract specific persons from two or more persons per images stored in the database, it is necessary to select major persons in an image. This study is concerned with human extraction by using the face features and its location in digital images.

View full abstract

In the image database, an algorithm for searching specific persons from similar images is useful to improve search efficiency. Therefore, this study proposes a method to extract three parts of the face: eyes, mouth, and hair, as the features to distinguish the person from others.

View full abstract

We propose group-based image retrieval method for video image annotation systems. Experimental results show that the proposed method can improve image retrieval accuracy to 90% up from the conventional method of 40%.

View full abstract

On a user generated video(UGV) sharing web site, huge amount of uploaded video materials makes it difficult to find target videos. We propose a method to create an index-video for quick searching of them, that consists of some short clips extracted from video materials by video analysis and prioritization technology.

View full abstract

Features of face images are analyzed by using the principal component analysis(PCA). It's possible to describe various facial impressions quantitatively by the combination of principal components. An individual face can be represented uniquely by the set of similarity measure with each impression. Retrieval of similar faces based on this method is discussed.

View full abstract

The difference between an input face and an average face is analyzed by using the independent component analysis(ICA). By enhancing the characteristic independent components for an individual input face, it's possible to generate facial caricatures automatically. Comparison between ICA and PCA(principal component analysis) is discussed by examining typical shape represented by each component.

View full abstract

Digital signage systems are getting popular nowadays. Although they can provide very expressive multimedia contents to audience, they just simply provide the scheduled contents. In order to add interactivity to the system, we've developed a new digital signage system which will display information proactively based on the results of face detection.

View full abstract

This paper discusses an evaluation method of the respiratory motion complexity of lung field. The prototype system for tracking feature points in a lung field and the experimental results have also been described.

View full abstract

The lung field motion model which represents a motion of each point in a lung field as linear combination of motion patterns is discussed. The method for pursuing shape of lung fields based on the model has also been proposed.

View full abstract

In order to achieve distance recognition to some objects in a landscape image by machine vision, we have been investigating a relationship between color sensation and the distance. In this study, variation of color parameters on images of an index panel was evaluated in various color coordinate systems. The report describes feasibility of the color variation.

View full abstract

In most images that can be roughly devided into two parts (object and background), focuses are on objects. Given the fact as prior knowledge, we propose a new robust way of automatically extracting objects from a image. Our method estimates a object mask and a background mask from a result of DWT, and performs graph cuts with the masks as seeds.

View full abstract

In this paper, we propose the technique that adaptive edge extraction and template matching method of face detection.

View full abstract

Homography is useful in image synthesis technologies. This paper describes a Homography estimation method for palm images by using color and shape features.

View full abstract

In telepresence, omnidirectional videos are often used for representing a real scene. However, omnidirectional cameras cannot usually capture the entire direction of view. In this study, an omnidirectional video without invisible area is generated by filling a missing region using an image inpainting technique for high realistic sensation in telepresence.

View full abstract

Texture-based image inpainting techniques have been widely used to remove undesired visual objects in an image. Conventional inpainting techniques used single patch for filling in a target region. However, different texture pattern often occurs as artifacts. In order to solve the problem; in this report, we propose a texture-based image inpainting method by superimposing weighted multiple patches.

View full abstract

In this paper, an irregular unsteadiness was achieved. Vibration was improved high restoration processing of high speed and the practicality was achieved by the choice adoption vector method that used movement vectors.

View full abstract

In this paper, since past degradation caused by flicker was improved by the one-dimensional histogram analysis method, it report. Since processing time is very fast, this technique has high practicality.

View full abstract

JPEG images are degraded by blocking artifact and mosquito noise caused by quantization on the block DCT domain, may be degraded by out-of-focus. The purpose of this paper is to make out-of-focus JPEG images to be sharp without enhancing the blocking artifacts and the mosquito noise.

View full abstract

In this paper, we investigate how the positions of scrambled regions affect on the image quality of scrambled images. Subjective assessment results show that scrambled region determination based on ROI (Region of Interest) has ability to control image quality.

View full abstract

We propose the method to directly superimpose a visible mark onto JPEG/MPEG picture code, by locally decoding the image code, superimposing the mark, re-encoding in INTRA code, replacing with original code, and matching between replaced blocks and surrounding blocks.

View full abstract

This paper focuses on performance improvement of intra prediction scheme. Intra prediction scheme of H.264 coding standard, which is the most representative scheme, has problem of prediction performance degradation due to block-based process. In this paper, the performance degradation is approximately modeled, and possibility of performance improvement by a novel prediction scheme using neighboring pixels is discussed.

View full abstract

This paper proposes the compression method based on hierarchical Block Truncation Coding for very low bit-rate. Using entropy as a criterion,this method properly detects the blocks having many colors, measured in the color space that matches human color perception, and divides them into small ones to improve the quality.

View full abstract

We previously proposed a novel video coding scheme based on iterative operations of motion compensation and matching pursuit(MP). In this paper, we improve coding efficiency of MP parameters by making use of edge information extracted from a motion compensated prediction image. Simulation results show that coding rates of the proposed scheme can be reduced by up-to 5%.

View full abstract

We propose a method for compression of motion image with context adaptation using cubic higher-order local auto correlation(CHLAC). The preliminary analysis indicates that this method is effective for compression of motion image.

View full abstract

Considering recent rapid increase of users on video-sharing sites such as YouTube^<TM> and illegal upload of copyrighted video clips, automated method for detecting pirated video clips is required. For this purpose, we have already proposed robust content-based copy detection scheme based on multiple color layout descriptors. In this paper, we evaluate robustness of our scheme against some transformations, and investigate the influence of the number of nearest neighbors for voting.

View full abstract

The authors study an objective method to estimate perceptual picture quality of the coded pictures based on No Reference framework. The proposed method extracts two image features that reflect the significance of blocking and flickering artifacts and estimates the subjective quality by weighted sum of the image features. Computer simulations show that the proposed method can estimate subjective quality at a correlation coefficient of 0.754.

View full abstract

This paper describes a method for robust gaze detection against distance changes between a person and cameras using stereo vision. The experiment showed that the gaze detection is accurate when the distance between the person and cameras is 50cm and 100cm.

View full abstract