Abstract. Polarization is a phenomenon that cannot be observed by the human eye, but it provides rich information regarding scenes. The proposed method estimates the surface normal of black specular objects through polarization analysis of reflected light. A unique surface normal cannot be determined from a polarization image observed from a single viewpoint; thus, we observe the object from multiple viewpoints. To analyze the polarization state of the reflected light at the corresponding points when observed from multiple viewpoints, the abstract shape is predetermined using a space carving technique. Unlike a conventional photometric stereo or multiview stereo, which cannot estimate the shape of a black specular object, the proposed method estimates the surface normal and three-dimensional coordinates of black specular objects via polarization analysis and space carving.

A plaster statue is an object created from a mold made from a stone statue, for example, and into which plaster is poured and hardened. The delicate shape is lost, however, during the molding process, and an issue arises in that the plaster statue has a smoother shape than does the original statue. Consequently in this research, by applying a sharpening filter to the 3D shape data of a plaster statue, highlighted contours comparable to those of the original stone statue are reconstructed for the 3D shape data. The objective is to prepare a stone statue that is used as a reference and to sharpen the input data to match the histogram of the edges of the 3D shape of the stone statue.

Range scanners based on camera-projector setups permit capturing dynamic objects at high frame rates at affordable costs, producing a time-varying set of points. This paper establishes a 4D media framework for acquiring, compressing and displaying these point clouds, albeit with more emphasis on the rendering aspects. Displaying these point clouds requires techniques capable of alleviating irregularities that are inherent in these low cost high-speed scanners. One of the contributions of this paper is a simple, general surfel splatting method that mitigates these imperfections. Rendering must also react quickly to the volatility of the 4D media dataset and cope with reasonably large quantities of points. To that end, this paper exploits modern features of the graphics hardware, namely geometry amplification and instancing, for efficient streaming and displaying of point clouds as either surfel or voxel primitives. A comparative performance analysis of the investigated strategies is also available.

We propose a method to observe cardiac beat from 3D shape information of body surface by using the active stereo with waved-grid pattern projection, and report preliminary experiments to evaluate validities of the proposed method. By comparing results of our method with those of electrocardiogram (ECG), we confirmed sufficient correspondences between peak intervals of depth changes between contiguous frames measured by the active stereo and R-R intervals measured by ECG. We proposed the visualization of the spatial distribution of depth change plotted on the 3D shape of chest surface. We confirm that the spatial phase difference, which is caused by heart pump ability, appears in the 3-D shape change of chest surface.

In this paper, we propose a novel 3-D endoscope system by using grid-based active stereo. In the proposed system, projection of a waved-grid pattern that consists of vertical and horizontal sinusoidal lines realizes accurate shape acquisition in sub-pixel accuracy. We develop a small pattern projector implementable to a head of a ready-made endoscope, and examine 3-D shape reconstruction by actual equipment. As the result of the measurement for a known-shaped object, which is a hexagonal cylinder, the error of length measurement is below 0.9% and the error of angle measurement is below 2.2%. We make a measurement of animal organ meat, and confirm that the system can reconstruct a 3-D shape of the organ surface.

Much research is in progress on the acquisition of high-density three-dimensional shapes by acquiring and combining shape data and normal data. The method proposed in this paper estimates normals to object surfaces by employing the photometric stereo method and combines the estimation with the three-dimensional shape acquired by a laser range sensor. Although the photometric stereo method presumes that light-source directions for each image are known, the proposed method uses its light-source direction estimates. By linearizing the images as preprocessing, specular reflection and shadow effects within the image are removed and the precision for the light-source direction estimations is increased. Since the proposed method does not require the light-source directions to be known, it offers the advantage of broad applicability for measurement work.

We propose a polarization-based method to enhance the visibility of an image by canceling the haze effect. Haze is a natural phenomenon that degrades the visibility of a scene. Aerosols in air reflect sunlight and cause polarization. Therefore, we analyze the polarization state of the observed light to remove the haze effect from a captured image. Our approach is to use two reference objects that are known a priori in estimating the parameters of the haze effect. Once the parameters are known, we can improve the image so that the scene is clearly visible. We also present an experimental result using a commercial polarization camera, which can obtain the polarization state of the scene.

Polarization is physical phenomena of the light which gives rich information of the scene even though it is undetectable by human eyes. In this paper, we analyze the polarization state of the light reflected on black specular objects in order to estimate the surface normal of the objects. Since polarization image of single view is not enough to uniquely determine the surface normal, we observe the object from multiple views. A rough estimate of the object geometry is obtained a priori by space carving so that the polarization state of the reflected light can be analyzed at the same surface point under multiple views. Unlike the photometric stereo or the multiview stereo which cannot estimate the surface normal and the geometry of black specular objects, the proposed method which is based on the polarization analysis and the space carving can estimate the surface normal and the geometry of black specular objects.

In the present paper, we propose a one-shot scanning system consisting of multiple projectors and cameras for dense entire shape acquisition of a moving object. One potential application of the proposed system is to capture a moving object at a high frame rate. Since the patterns used for one-shot scanning are usually complicated, and the patterns interfere with each other if they are projected onto the same object, it is difficult to use multiple sets of patterns for entire shape acquisition. In addition, the overlapped areas of each object have gaps and errors are accumulated. As such, merged shapes are usually noisy and inconsistent. In order to address this problem, we propose a one-shot shape reconstruction method using a projector to project a static pattern of parallel lines of one or two colors. Since each projector projects only parallel lines with a small number of colors, these patterns are easily decomposed and detected even if the patterns are projected multiple times onto the same object. We also propose a multi-view reconstruction algorithm for the projector-camera system. In the experiment, we built a system consisting of six projectors and six cameras, and dense shapes of entire objects were successfully reconstructed.

The human brain recognizes pictures that are first obtained by the photosensitive cells in the retina. Color is a visual perception composed by the stimulus of three kinds of photoreceptors called L, M, and S cones. The human eye often recognizes different spectral distribution as same color since each cone has wide spectral response. This phenomenon is known as metamerism. Our research project aims to innovate a novel form of artistic illusion by fully making use of metamerism. This paper proposes a method which estimates the mixture ratio of paints that can cause metamerism.

Recently 3D scanning systems are capable of modeling entire dense shapes that evolve over time with a single scan ( a.k.a. one-shot scan). In particular, structured-light-based systems have emerged as one-shot shape reconstruction methods that project a static grid pattern onto the object surface. This pattern allows the scanning of moving objects while still maintaining dense reconstruction. One-shot scanning systems are then capable of producing 3D+t ( a.k.a. 4D) spatio-temporal models with millions of points. As a consequence, effective 4D geometry compression schemes are required to face the need to store or transmit the huge amount of data, in addition to classical static 3D data. In this paper, we propose a 4D spatiotemporal rate-distortion (RD) optimized point cloud encoder via a curve-based representation of the point cloud, particularly well-suited for one-shot scanning systems. The object surface is naturally sampled in a series of curves, due to the grid pattern. This motivates our choice to leverage a curve-based representation to remove the spatial and temporal correlation of the sampled point along the scanning directions through a competitive-based predictive encoder that includes different spatio-temporal prediction modes through an RD cost computation control. Experimental results show the significant gain obtained with the proposed method.

Recently it is relatively easy to produce digital point sampled 3D geometric models. In sight of the increasing capability of 3D scanning systems to produce models with millions of points, compression efficiency is of paramount importance. In this paper, we propose a novel competition-based predictive method for single-rate compression of 3D models represented as point cloud. In particular we aim at 3D scanning methods based on grid pattern. The proposed method takes advantage of the pattern characteristic made of vertical and horizontal lines, by assuming that the object surface is sampled in curve of points. We then designed and implemented a predictive coder driven by this curve-based point representation. Novel prediction techniques are specifically designed for a curve-based cloud of points, and been competing between them to achieve high quality 3D reconstruction. Experimental results demonstrate the effectiveness of the proposed method.

The estimated height of the mushroom cloud that formed after the A-bomb explosion at Hiroshima has been a controversial issue for many years. In this work, we have attempted to measure the cloud height from existing photos taken at the time 10 from airplanes and from the ground. First, we determined the precise locations where the pictures were taken. Next, we approximated the three dimensional shape of the cloud using multiple spheroids. Finally, we obtained measurements of the height and width of the mushroom cloud from the approximated cloud shape. Our experimental results suggested that the height reached a maximum of about 16 km. 15

With the increasing demands for highly detailed 3D data, dynamic scanning systems are capable of producing 3D+t (a.k.a. 4D) spatio-temporal models with millions of points recently. As a consequence, effective 4D geometry compression schemes are required to face the need to store/transmit the huge amount of data, in addition to classical static 3D data. In this paper, we propose a 4D spatio-temporal point cloud encoder via a curve-based representation of the point cloud, particularly well-suited for dynamic structured-light-based scanning systems, wherein a grid pattern is projected onto the surface object. The object surface is then naturally sampled in a series of curves, due to the grid pattern. This motivates our choice to leverage a curve-based representation to remove the spatial and temporal correlation of the sampled point along the scanning directions through a competitive-based predictive encoder that includes different spatio-temporal prediction modes. Experimental results show the significant gain obtained with the proposed method.

The paper folding as known as "origami" is the Japanese traditional folkcraft that everyone can enjoy. Recently, several researchers are dedicated to virtually representing the origami in personal computer. However, previous methods less reflect the uniqueness of the actual origami folded by people. In this project, we build a software in order to represent the 3D model of origami crane which has the same message on its surface drawn by the user who folded it in the real world.