The expectations for Electronic Paper can be summarized as readability, compactness, multi-functionality, and resource conservation. Paper-like readability should be accepted as the most important target of Electronic Paper, when we consider that no existing electronic display is as comfortable, to read, as paper. A concept of readable Electronic Paper is proposed, and its expected properties, applications, and various candidate technologies are reviewed. Several experiments to clarify how to realize ideal readability on electronic medium are also introduced. It is suggested that free handling of medium and total view for each page without scrolling are essential factors for getting paper-like readability.

View full abstract

Retail shops and sports clubs introduce 3-D body scanners in order to use 3-D body images for fashion and healthcare services. A large amount of 3-D body images obtained by these systems are stored in these shops and clubs. Technologies for browsing and reusing 3-D images based on digital human modeling are required. Furthermore, quality control of distributed systems is also required. State of the art technologies for storing, browsing, and reusing 3-D body images are mentioned.

View full abstract

Many consumer digital color cameras have a single image sensor with a color filter array. The data captured by the single image sensor are called raw data. An effective compression of the raw data is highly demanded. This paper proposes a raw data compression method using existing image coding framework. The proposed coding method is performed to minimize error between the observed raw data and the decoded raw data. Experimental comparisons demonstrate that the proposed method has high performance compared to existing methods.

View full abstract

We have developed an efficient BTF compression and reproducing technique based on a dichromatic reflection model. For the reproduction of the gloss appearance of rough surface, Bi-directional Texture Function (BTF) rendering is thought to be used in this system. However, efficient compression and reproduction methods are required for real-time rendering. In this paper, we propose an efficient approach of BTF rendering by separating the reflected light to the specular, the cast shadow and the invariant diffuse elements. The efficient compression for the BTF rendering is performed by applying the Local Principal Component Analysis (LPCA) to the specular and the cast shadow images, since their appearance depends on the direction of the viewing or illumination. This method performs accurate reproduction by fewer principal components in comparison with the conventional method. The final projection image in an arbitrary direction of the viewing and illumination is synthesized from the specular reflection image, the cast shadow image and the invariant diffuse reflection image. Finally, the proposed method is implemented on the appearance-based display, and demonstrated high-speed BTF rendering which is two times faster than the conventional method.

View full abstract

Since JPEG 2000 is selected as the standard of Digital Cinema, it attracts a great deal of attention. However, JPEG 2000 coding requires huge amount of computational complexity for a high definition image with more than 200 million pixels required in Digital Cinema standard. Especially, EBCOT is a bottleneck because of its difficulty in parallel processing. This paper proposes a parallel coefficient bit modeling (CBM) algorithm which can process a bit plane, a stripe and a pass in parallel and a CBM architecture based on it. Hardware evaluation results show that a 733k Gate (TSMC 0.18 μm) circuit can process the code block (block size 32×32, 30bit planes) which satisfies Digital Cinema standard in real time.

View full abstract

The Internet has dramatically increased the number of its user in exchanging image data. The amount of image data is huge compared with that of text data. Even under these conditions, users would like to recognize the images as early as possible. Previously, Internet users have used Progressive JPEG which can be seen an approximation to the whole image very quickly, with gradual improvement of quality as one waits longer. However, the order of data transmission based on DCT coefficients has not been defined. As a result, it has caused problems that an unnatural improvement of displaying the image occurs depending on network conditions. In this paper, we propose a variable frequency band division, and apply our method to DWT coding for mobile devices, such as cellular phones and PDAs. Additionally, we attempt evaluation experiments for the methods availability.

View full abstract

Watermark method is proposed for improvement of data hiding performance. Digital watermarking in MRA: Multi Resolution Analysis is used abundantly. It is the method of embedding restricted data at the specific frequency component after wavelet decomposition, and since restricted data is also discoverable by exploring at what frequency component it is embedding, the subject is left behind to privacy. The method of giving eigenvalue decomposition as this preprocessing is proposed. Since only the author who holds the original picture image can know the eigenvalue of an original picture image, it is only an author to restore this. However, if this method permits a certain amount of error using the circulation picture embedding restricted data, since it can presume the approximate eigenvalue, its privacy is inadequate. Here, the preprocessing which gives oblique-coordinates conversion after eigenvalue decomposition is proposed further. Since the degree of slanting crossing angle of oblique coordinates can be set up arbitrarily, privacy improves. Thus, digital watermarking to propose is excellent in privacy. The effect was checked using the picture by which this is used abundantly as a standard picture of data compression. It is confirmed that data hiding performance of the proposed method is almost double compared to the conventional method based on MRA.

View full abstract

The Retinex is a human perception-based image processing algorithm which provides color constancy and dynamic range compression. A real-time processing of the Retinex is required to meet the needs of various potential users. The Retinex processing contains a relatively large number of complex computations, thus to achieve real-time performance using current technologies requires specialized hardware and software. However, there is a limit to the image size that the Retinex can output an improved image in a short time even if any machines are used. As one of the solutions by approaching from software implementation, in this paper, we propose a serial Retinex algorithm for improving appearance of spatially localized images. Our algorithm does not output a whole image by the Retinex at a time, but the algorithm calculates it consecutively and output it sequentially. So, with progress of processing time, the appearance will be gradually improved and the processing region will be gradually spread out. We verified the effectiveness of the proposed algorithm by experiments.

View full abstract

This paper proposes an interpolation algorithm and LSI architecture for a camera with a fish-eye lens. Since the fish-eye lens has around 180 degree field of view and a deep depth of field, it's very promising for a surveillance camera, a video conference system, and so on. Combination of a cubic spline interpolation and a direct interpolation from a RAW image enable to obtain high quality image. A three-stage linear interpolation scheme and a dedicate operators boost up processing performance. Evaluation results base on a subjective test show that our proposed algorithm can generate high-quality interpolation image. Furthermore, a dedicated image processing core with 290k gates can process HDTV (1920×1080, 60fps) image by using 0.18 μm CMOS technology.

View full abstract

This paper describes a method and experiment on online handprinted “Hiragana” character recognition. The recognition process is composed of a rough classification step by Procrustes analysis, which is a generalization of linear regression model, and a detailed recognition step using vector correlation. The Procrustes analysis is a permissive one for rotation changes. And vector correlation is a excellent measure for acquiring directional information about local strokes. The proposed method is corroborated by a computer simulation, which correctly classified 99.2% of 2300 patterns in 46 categories.

View full abstract

Contents of sports tickers in broadcasted sports videos could be efficient as queries in analyzing the video contents. However, since sports tickers are displayed with various colors and brightness, it is difficult to recognize inner objects of the sports tickers only according to the brightness distribution. Besides, in order to recognize what the objects represent, it is indispensable to investigate the change of these objects according to time traveling. In this paper, the proposed method can recognize locations of the out ticker, score tickers, and runner markers in the sports ticker of broadcasted baseball videos analyzing the brightness histograms obtained from every of grayscale and color images, and the change of the sports ticker with the correlation coefficient. Experimental results with four broadcasted baseball videos have shown that the proposed method can exactly work in the case when the out ticker is represented by a letter.

View full abstract

Contour detection is a fundamental step to scene analysis and interpretation. However, because contours often locate in rich texture background it is still a difficult task in realistic vision. Through multi-scale analysis, it becomes clear that edge responses of real object contours are relatively stable across scales, while those from noise or texture background are not. In this paper, a salient contour detection method is proposed based on the scale invariance of piecewise linear approximation of real object contours. Firstly, an image pyramid is efficiently constructed by repeatedly smoothing and sub-sampling the image. Secondly, the piecewise linear approximation of contours in multiple scales are extracted and then a collinear line grouping process is implemented to improve the connectivity of the contours. Thirdly, the new salient line segments are generated based on the analysis on the stability of line segments across scales. Experimental results show that the proposed method can effectively improve the connectivity and saliency of the contour detection compared with the former method.

View full abstract

In the field of meteorology, the analysis of satellite images is important to understand current and future atmospheric states, as the satellite images have advantages on range and frequency of the observations. In this study, we introduce a system of cloud classification and detection of cirrus streak for comprehensive and structural understandings of atmospheric states. In the cloud classification, the proposed system classifies clouds into appropriate cloud classes using the past cloud classes and cloud velocity based on the strong correlation of successive satellite images. Moreover, our system detects cirrus streaks using cloud shape, velocity and cloud classes obtained by the cloud classification. In the experiments, the effectiveness of the proposed system is investigated with actual satellite images.

View full abstract

Real-time defect classification on wafer inspection is important for semiconductor manufacturing. Rule-based classification is popularly employed because of its advantages that classification rules can be decided based on theories and experiences and can be easily understood by anyone. But it is difficult for non-experts to decide them. So we propose automatic rule generation algorithm based on teaching of feature data with correct defect classes. The proposed algorithm constructs a decision tree based on separability measures, which are defined using entropy and c-squared statistic. The algorithm was evaluated using a set of feature data obtained with a wafer inspection tool. The data set included 186 defects of six classes and its classification accuracy was 86.6%. This result shows the algorithm works usefully when features are appropriate for classification.

View full abstract

This paper proposes a technique for high speed recognition of human motions. For an intelligent robot which will help human lives in various ways, it is indispensable to recognize human motions or behaviors irrespective of its observing direction. For this purpose, we propose a structured motion database provided by an eigenspace having a B-tree structure within it. In the motion database, every appearance of a human motion is to be registered. Another idea for speeding up the recognition is to describe a motion in a compressed form. We employ a JK method to represent a motion sequence in a single image. Performed experiment showed satisfactory results.

View full abstract

In this paper, we propose a method of surface details representation based on Level-of-Detail (LOD) of haptic feature. Previous studies did not care for LOD control of surface details: Surface details were generally compressed collision detection and their details were lost. To solve this problem, we propose a method of compressing surface details. Our method realizes two rendering methods: the rendering based on conservation of structure features and the rendering based on selection of partial shapes.

View full abstract

In this paper, we propose the image analysis for growth development of rice leaves movement. We propose the technique of imaging for growth of rice and of the visualization for growth of the leaf using temporal-spatial analysis and apply these techniques to growth analysis. To be established the visualization technique for leaf movement, we corrected image series of rice growing during some month and processed acquired image series coordinate conversion and projection space-time map. Observation of the physiological phenotypes in the growth of rice, such as change of the movement pattern, circumference, and growth movement, can be performed using the visualized result.

View full abstract

This paper proposes a Multiple Messages Embedding (MME) scheme for achieving plausible deniability with DCT-based Mod4 steganography. To implement MME, we utilize the structural feature of vGQC (Valid Group of 2×2 adjacent quantized DCT Coefficients) in Mod4 as message carrier. vGQC's can be partitioned into several disjoint sets by differentiating the parameters where each set could serve as an individual secret communication channel. A maximum number of 14 independent messages can be embedded into a cover image without interfering one message and another. We can generate stego images with image quality no worse than the conventional Mod4 when embedding at the same rate. Results for blind steganalysis are also shown.

View full abstract

In this paper, we propose a method to improve the performance of Super-Resolution for video sequence that contains moving objects by realizing Dense Motion Estimation (DME) virtually. The performance of Super-Resolution heavily depends on the accuracy of motion estimation. Therefore, if accurate DME that allocates motion vector to each pixel can be achieved, the performance of super-resolution must be remarkably improved. However, it is difficult to realize pixel-level DME accurately on the enlarged images, because observed images lost information necessary for motion estimation via degradation process. In this paper, we first apply DME using Hierarchical Block Matching (HBM), which is one of the conventional methods, to Super-Resolution, and verify that the accuracy of DME using HBM depends on the quality of observed images. Precisely, DME using HBM works effective only for high quality observed images, but the effectiveness seriously deteriorates for low quality observed images. Then, we propose a method that realizes virtual DME using Overlapped Block Matching (OBM), which is often used to reduce block noise in motion picture coding. While this method does not improve the accuracy of individual motion vector, it apparently increases the number of reference frames by allocating multiple motion vectors to a small region, which improves the performance of Super-Resolution. Through computer simulation, we show that the proposed virtual DME is robustly effective for Super-Resolution irrelevant to the image quality of observed images.

View full abstract

In this paper, a new technique aimed at face posture estimation in 2D images is proposed. Posture estimation is an important process, which is used to preprocess face recognition. However, the conventional technique had a problem in that a gap of 10deg or more is caused when taking a picture with the lighting brightness changed according to the place and time. The proposed technique corresponds when the lighting change is used such that the positive and negative signs of the radiance value are reported to be strong compared with the lighting change. Moreover, our technique aims at a posture estimate within error 5deg. Therefore, we use 3D shape data of the face so that a multi-angle turn is possible. As a result, the proposed technique obtains excellent results of about 87% for the input-changed bright image and the success rate of the posture estimation by the technique becomes about 47% by CSU+LDA of the conventional technique.

View full abstract

Modular Visualization Environments (MVEs) are commonly used in many disciplines including fluid science, because of their archivability and extendability, which domain-specific visualization software does not possess. However, it is still difficult for scientists and engineers, who do not have to be visualization experts, to fully utilize a bunch of visualization modules by taking into account their pros&cons and usage for constructing optimal MVE applications. In this paper, we propose a system, called GADGET/FV, which assists the users in designing their MVE applications for flow visualization problems by relying on an extended Wehrend Matrix as its goal-oriented taxonomy to retrieve effective visualization techniques to meet their goals and requirements. In the GADGET/FV system, adaptive and detailed user support in accordance with the progress of their visual analysis is realized through several feature mechanisms to save/reuse visualization processes, and to link research memos explicitly to corresponding visualization results. Practical fluid analysis is performed to illustrate the usefulness of the present system, with a proof of the system capability in reference to Robertson's criteria to evaluate the performance of visualization systems.

View full abstract

This paper introduces an implementation method of “awareness video communication system” based on the JPEG2000 (JP2K) international standard for digital image coding. The system can display “awareness” of people by transmitting minimum data necessary for displaying the awareness. The minimum data contains (1) pixels in person region, (2) higher bit-planes for displaying semi-transparent person and (3) a part of band signals for blurred person region signal. It becomes possible to reduce redundancy of data transmission since (2) and (3) above are implemented by the “bit-plane decomposition” and “wavelet transform” of the JP2K respectively.

View full abstract