Image-based lighting is widely used to represent real-world, complex lighting in recent years. Image-based lighting captures surrounding environment illumination as an environment map, and uses it for the illumination. Although realistic images can be rendered by using image-based lighting, the computational cost using image-based lighting can be prohibitive since rendering under image-based lighting requires the product integral of incident radiance, visibility function, BRDF over the hemisphere. This paper surveys the recent papers on fast rendering using image-based lighting.

View full abstract

Global illumination is indispensable for realistic image synthesis in computer graphics. In global illumination computation, we account for not only the direct lighting from light sources, but also the indirect lighting from reflected or refracted surfaces. The direct and indirect light transports are altogether mathematically formulated via the rendering equation. In this article, we focus on reviewing the rendering equation and its fundamental solutions, such as the Monte Carlo methods and the Markov Chain Monte Carlo methods. In addition, we describe the recent advancement in global illumination computation.

View full abstract

The environments where robots are used are changing from orderly environments such as factories to diverse environments such as human living spaces. For robots to move autonomously in the environment, they require environmental maps composed of geometric features. However, it is difficult to prepare environmental maps in advance for diverse environments such as human living spaces; therefore, it is necessary to develop a method enabling robots to construct environmental maps autonomously. Recently, the use of a low–cost, user–friendly RGB—D sensor, such as Kinect, has attracted attention. However, because an RGB—D sensor is primarily used to record human movement, it is not suitable for highly precise measurement. Therefore, the accuracy of depth information is not high, and there are limitations when using the sensor to measure detailed shapes accurately. In this paper, we propose a sensor calibration method to acquire highly precise range–imaging data. In our proposed method, linear functions are employed for correction purposes by applying the method of least squares to each sensor and each pixel. Furthermore, we demonstrated that it is feasible to perform highly precise corrections of the point cloud data obtained from the RGB—D sensor by utilizing two types of correction formulas for short–range distances and long–range distances depending on the distance from the sensor.

View full abstract

Many kinds of switching-type denoising filters have been proposed to detect and denoise noisy pixels in an image effectively. Our multi-directional switching median filter uses a 2×2 pixels noise detector and an averaging method. To discriminate noisy pixels in the image, our filter requires a threshold like other switching-type filters. The optimal value of the threshold varies depending on the strength and quantity of the edge of the image, which means the optimal threshold varies from area to area of the same image. Focusing on this point, we propose a new method. It divides the image into segments and, for each of them, sets a threshold proportional to the average edge strength of the segment. Our experiment verified that this method could improve image quality.

View full abstract

Parallel coordinates is well-known as a popular tool for visualizing the underlying relationships among variables in high-dimensional datasets. This visualization technique is useful for visually understanding the degree of correlation between data samples in terms of two adjacent axes. However, this representation still suffers from distracting visual clutter especially when the numbers of data samples and their associated dimension become high, because the associated polyline samples intricately overlap with each other within the limited screen space. This paper presents a method of alleviating such visual clutter by contracting multiple axes through the analysis of correlation between every pair of variables. In this method, we first define the similarity between a pair of dimensions as the value of the correlation coefficient, and construct a subgraph from the complete graph through eliminating all the edges in that their absolute correlation coefficients are less than some threshold, and then reorder the multiple axes by projecting the nodes onto the primary axis obtained using the spectral graph analysis. This allows us to compose a dendrogram tree by recursively merging a pair of the closest axes one by one. Smooth animation of the associated axis contraction and expansion has also been implemented to enhance the visual readability of behavior inherent in the given high-dimensional datasets. We also conducted a user study to investigate how the correlations among coordinate axes are better visualized using our approach.

View full abstract

Non-photorealistic rendering (NPR) techniques have been proposed for generating paper-mosaic images composed of regularly-formed paper-pieces such as ellipses or squares. Hand-made paper-mosaics, however, consist of various form of pieces. We present, in this letter, an NPR method for generating such paper-mosaic images composed of paper-pieces of various forms and visual effects by growing lines from points placed with the Poisson disk sampling (PDS) for tessellating an image into cells. Points are placed almost equidistantly by a radius-ordering PDS. The radius is shrunk around edges for preserving object boundaries. Images are tessellated by growing lines along contours of the luminance of each pixel. Extended examples are also shown with addition of gradation or hatching to every colored cell.

View full abstract

Although the arithmetic coding method is known as a highly effective entropy code, its encoding and decoding are time consuming due to the symbol-by-symbol arithmetic operations. Several approaches have been proposed to reduce the computation time by replacing the multiplication for the augend calculation with the addition, subtraction or shift operation, or by using some conditions in the state transition table for probability estimation. However, these approximations cause a loss of the coding efficiency. The recent progress of the hardware technology allowed us to use faster multipliers and bigger storage media. Therefore, it is important to consider to speed up the arithmetic coding with the recent technological environment. In this paper, we propose a fast binary arithmetic coding by using the binary source expansion and precomputed probability tables of expanded symbols. Simulation results show that the encoding process and decoding process in the proposed method is faster than that in the conventional multiplication-based binary arithmetic coding method while keeping the coding efficiency.

View full abstract

In this paper, we propose a novel image segmentation method exploiting an image abstraction method aiming for the application of image transmission and object-based image compression under very narrow bandwidth. Image abstraction is a method which reduces the complexity of a photograph and changes it into a picture like an artwork. In addition, our proposed method utilizes distance information obtained by a stereo camera and plane fitting. We conducted experiments using images obtained by two distinct stereo cameras and Inria 3DMovie Dataset v1, and subjectively confirmed that the proposed method is effective and promising.

View full abstract

In the template matching by using Particle Swarm Optimization (PSO), it is a waste of time to evaluate the similarity at the same searching region after repeating update. This paper proposes ActivePSO template matching that skips evaluation if a searching region is not needed to evaluate in template matching by using PSO. To decide if it is needed to evaluate the searching region, the upper limit of the similarity is obtained from the evaluation history. Therefore, processing time could be shorter than template matching by using PSO. In our experiments, proposed method was compared with current template matching by using PSO, and confirmed that processing speed was 3 times faster than current method without performance degradation.

View full abstract

This paper presents features for measuring the congestive extent as a means for diagnosing internal hemorrhoids using endoscopic images. A computer-aided diagnosis for the hemorrhoids could be effective for young or even general practitioners as a second opinion because diagnosis of hemorrhoids needs much experience. Considering characteristics of the hemorrhoids, the proposed method measures degree of the congestion in the images and the size of the congestive regions as abnormalities of the hemorrhoids. Experimental results of the discrimination between normal and abnormal cases using the proposed abnormalities for 204 images (96 normal cases and 108 abnormal cases) have shown that the abnormalities are well effective to measure degree of the congestion in internal hemorrhoids.

View full abstract

This paper presents a method to synthesize a high speed soft body character animation in real time. Especially, not only primary but also secondary physics based deformation in an under skin fat structure driven by a skeletal motion are considered. However, as a high fidelity soft body animation method, high cost FEM based methods are usually introduced, the method which represents a high speed and robust soft body animation using simplified elastic simulation method was proposed. In this method, they applied skinning result to fat and skin, and then a vibration was limited to be a very small and unnatural. So in this paper, we proposed a method to divide skinning and simulation by modeling an inside structure under skin approximately and automatically. As a result, by controlling parameters such as volume and elastic parameter at each layer freely. Consequently, by a user interaction with parameter adjustment, a definition of soft body material according to the location can be implemented to generate and edit a proper soft body motion. After an evaluation of soft body motion synthesis based on a several character models, this method is approved to be very effective to make a character animation with soft body characters.

View full abstract

Thanks to the recent progress of information technologies, the role of geometric patterns as communication media has been increasing. Barcodes have been widely used for computers to read information from the printed documents. In this paper, a communication method through embedded information in a geometric pattern composed of simple graphical tiles is proposed. Interesting structure of the pattern would possibly make it subjectively less obstructive when the pattern is printed on a material. A simple decoding method is given by deliberate use of geometric features of the pattern. Then the method is applied to document authentication. Computer simulations show that the information can be retrievable from scanned digital data obtained using consumer equipment for printing and scanning.

View full abstract

In this paper, we propose Lip-sync personal authentication system based on the movement features of the lip. Lip-sync means the utterance action of the silence. The user using this system is possible the personal authentication by carrying out the utterance action of the silence toward the camera. The proposed method has the practical advantage by using only the image information without the audio information of the mouth, for examples; the incapacitation of the acoustic environment noise, leakage of the authentication keyword can be prevented, the correspondence to the user with the difficult utterance. In this proposal technique, six time series data acquired by the characteristic tracking of the lip are used as behavioral features. Acquired time series data are compared by the dynamic programming matching method every one dimension. The score which integrated calculate six distances is used to the personal authentication. The effectiveness of proposal technique was examined by the personal authentication experiment. In addition, the effectiveness of each features was examined by gaining the weight to each features.

View full abstract