digital images are becoming richer, higher-reality and three-dimensional. There, a signal-to-noise ratio is no longer enough. In this paper, "Reality Measure" is newly proposed as the overall criteria. It is defined as a function of Immersion, Picture Quality and Three-Dimensional Sensation. With this measure, several 3D displays are evaluated and compared with the subjective test results.
In asynchronous e-learning system, quality of its contents is extremely important. We suppose that the contents provide lecture videos which show a lecture slide and a teacher pointing the slide with the mouse pointer on the computer screen. Although information of where the teacher is pointing at is indispensable to e-learning audiences, lecture videos are often compressed, which degrades image quality and make it difficult to recognize the position of the pointer. To improve reproduced image quality of the contents, we have developed a pointer tracking application which logs history of mouse movements and developed an efficient scheme to reproduce a distinct mouse pointer image on the screen by using the history logs. A preliminary experimental result indicates that the scheme efficiently improves reproduced images.
This report addresses a Japanese sign language dictionary system that is accessible by mobile phones through telecommunication networks. The system has two main retrieving methods: description parameter method and Japanese gloss method. We also realised another modified version of the parameter method called sequential method. This method shows the meaning of each parameter to be specified. The system has the function of showing text description for signing motion picture so that a user can recognise signing easily.
This study was aimed at developing a novel face recognition scheme based on human subjective judgment. The proposed scheme achieved higher precision than conventional face recognition schemes used for the similar face image retrieval. In order to obtain solid improvements from subjective judgment, the proposed scheme uses difference images between pairs of similar face images which resemble each other. These difference images are only composed of unnecessary elements when human beings evaluate similarity between face images. Then, the proposed scheme can remove these unnecessary elements from the original face image using projection onto a feature space of difference images, and can leave only important elements for judging the similarity. Simulation results showed the proposed scheme improved ranking output of similar face image retrieval approximately 28.6 percent more than conventional schemes.
We propose a lossless compression method of multi-level images using a Markov model coding. In Markov model coding, the number of the contexts increases exponentially depending on the numbers of reference pixels and the gray levels of the image. Since the more the number of states increases, the less the number of pixels per a state decreases, Markov model entropy does not indicate the ideal code length of the coding model. In this paper, at first, we propose an integrating method of the context in Markov model coding. Secondly, we propose a combined method of an integrated-contexts Markov model and a predictive model for the multi-level images.
The simulation results show that our proposed method can reduce effectively the average code length compared with the conventional coding method.
In coding of color images, R, G and B signals are converted into luminance Y and chrominance Cr, and Cb components. In the lossless mode, the reversible component transform method has to be introduced. Some reversible component transform methods, such as the method in JPEG2000 or FRExt in H.264, have been proposed. In this study we consider the coding performance of these reversible component transform methods. Moreover we propose a new reversible component transform method which approximates Y signal more preciously. We report that the comparison result of our proposed method to the conventional method.
H.264 FR-Ext, one of the standard video coding scheme, can encode over HD size resolution video since applicable maximum resolution is defined as 4096×2304. However, in case encoding super high definition video, such as 4K system, the coding performance may improve by extending the motion compensation estimation size since the size used by the scheme is too small compared with the image resolution. Therefore, the method to improve H.264 FR-Ext coding performance with super high definition video by extending macroblock size is proposed in this paper. The effectiveness of the proposed method is evaluated by several experimental results.
MPmark is a digital watermark technology for MPEG videos. The feature of MPmark is that it inserts watermarking in the uncompressed video, and retrieves the digital watermark directly from MPEG compressed video data. We propose a method of retrieving the digital watermark from decoded videos. The effectiveness of this method is examined by experiments. As a result, it is shown that the proposal method can retrieve from decoded video the digital watermark with high probability.
Recently, a various digital watermark methods are proposed to prevent unauthorized copying and the falsification of digital data. In this paper, we propose a new digital watermark method using the vectorization of the wavelet transform coefficients. In our method, a vector in which the 1 bit of watermark image is embedded is composed of three coefficients of the lowest band in the wavelet domain. Embedding is performed by changing the Euclidean distance between the original vector and the basis vector. In our proposed method embedding without visible degradation of the image quality can be performed by using the characteristic of the vector distribution. Moreover, the robustness simulation for image processing such as JPEG compression is performed.
Telecommunications industries have been undergoing remarkable market changes for the past 100 years, from the public utilities businesses under the natural monopoly to the global competitive environments. The changes have been based on emerged and emerging new technologies, which make it possible to open the regulation barrier for new services. Some interesting factors of these changes will be introduced and discussed, including some recent topics of telecommunications markets.