[in Japanese]

View full abstract

We propose a new region extraction and image composition method with chromakey using a two-tone striped background and active contour models.A problem with using conventional chromakey techniques is that if objects have the same color,they appear to be part of the background.To improve the clarity of the image,we utilized adjacency conditions between two-tone striped areas on the background,and extracted foreground regions whose colors are same as those of the background.In addition,by using active contour models for the extraction result,we significantly improved the accuracy of the extraction.The validity of our method was shown through experiments.

View full abstract

We need to project coherent laser beams onto an object in order to produce a hologram for a real object.However,this requirement restricts the shooting arrangement in electronic holography.Using integral photography(IP) enables a hologram to be created from a reconstructed IP image,and allows us to electronically display a 3-D image of an object photographed under natural light.We propose a method that employs fast Fourier transform(FFT) to calculate optical transmission during hologram shooting and that requires only half the calculation load required by the conventional method,i.e.,that based on the Fresnel diffraction.We used this method to produce a hologram from an integral photography image of an object photographed under natural light.We then used a liquid crystal device to optically reconstruct a holographic image of the object and evaluated the effectiveness of the method.

View full abstract

The accommodation mechanism has been experimentally shown to function as a demodulator in perceiving digital-halftone still images. Stimuli used in the experiments consisted of digital-halftone sine-waves and numerical figures expressed by an error-diffusion method. The accommodative responses were measured with an infrared optometer while the subjects perceived the dot component or the signal component modulated by the dots. The following results were obtained: (1) the accommodation lag changes depending on which component is perceived,and increases more for the signal component perception than the dot component; (2) the accommodation lag when perceiving the signal component increases as the sine-wave's spatial frequency decreases and the figure size enlarges; (3) as long as the viewing distance is over 1 diopter,it does not affect the above results. These results suggest that the change of accommodation lag,which corresponds to the bandwidth of low-pass filters,affects how digital-halftone images are perceived,i.e.,that the accommodation mechanism can also play the role of the demodulator in human vision. Moreover,we clarified that the resting state of accommodation does not affect the accommodative responses to digital-halftone images.

View full abstract

Multiple-camera systems are emerging gradually.In order to let audiences feel comfortable when cameras are switched or when a free viewpoint video generated is shown,colour correction between cameras is necessary.Colour correction has usually two steps; the first step is to adjust camera parameters such as gain,brightness and aperture before capturing,and the second step is to modify captured videos by image processing.This paper deals with the latter step,which does not need a colour pattern board to be prepared,uses SIFT(Scale Invariant Feature Transform) to detect correspondences for occlusion handling,treats RGB channels independently,and transforms the distorted videos with a non-linear function.Our experimental results show that there was a better correlation between the corrected videos than that between captured ones.

View full abstract

A 3D human pose is estimated from a monocular image using a retrieval-combination approach that exploits the broad capability of example-based approaches and the flexibility of parts-based approaches.Instead of storing and searching for similar full-body examples,we adopt a half-body representation (i.e.,either upper-body or lower-body) to reduce a large full-body database into a compact half-body database.The database can create millions of poses by valid half-body combinations.For a given query image,half-body candidates are first retrieved from the database by partial-shape matching.Valid half-body combinations of these candidates are selected based on a learned combination constraint,and then the optimal combination(s) is(are) chosen in a coarse-to-fine evaluation method.We show good experimental results for estimating poses with both synthetic and real images.Our approach has less time and space complexities than example-based approaches and ensures more realistic 3D pose estimates than those of parts-based approaches.

View full abstract

We present a method that uses 3D shape reconstruction and texture mapping to improve the image quality of free-viewpoint video.Three-dimensional shape refinement is used to represent 3D object shapes more accurately without increasing the processing time or the required memory space. We describe how our video generation method uses 3D shape smoothing and voxel space deformation and show some experimental results.

View full abstract