Kei Utsugi

Seam carving for stereo images

We propose a content-preserving stereo image editing technique by using stereo extension of seam carving. Seam carving is the process of deleting or duplicating connected paths, or seams, that consist of less important pixels in order to resize the image while preserving the image content as much as possible. The problem targeted in this paper is how to apply the seam carving method to a pair of stereo images, where the consistency between the left and right images should also be preserved through the seam carving process. For this consistency, we introduce new energy terms. Based on these energy terms, seam pairs between the input images are classified into two types, corresponding seams in order to maintain consistency or occluded seams to change the consistency intentionally. The novelty of this paper is that stereo matching results are fused into the framework in the seam carving. We demonstrate that by selecting seams in an appropriate way we can virtually manipulate the depths of objects in the scene.

GO-STOP Control Using Optical Brain-Computer Interface during Calculation Task

We have developed a prototype optical brain-computer interface (BCI) system that can be used by an operator to manipulate external, electrically controlled equipment. Our optical BCI uses near-infrared spectroscopy and functions as a compact, practical, unrestrictive, non-invasive brain-switch. The optical BCI system measured spatiotemporal changes in the hemoglobin concentrations in the blood flow of a subjects prefrontal cortex at 22 measurement points. An exponential moving average (EMA) filter was applied to the data, and then their weighted sum with a taskrelated parameter derived from a pretest is utilized for time-indicated control (GO-STOP) of an external object. In experiments using untrained subjects, the system achieved control patterns within an accuracy of ±6 sec for more than 80% control.

Proportional constraint for seam carving

Seam carving is an image processing operator for content-aware image resizing [Avidan and Shamir 2007]. It generates an energy map from gradient intensity of pixels and searches for seams, which are vertical or horizontal continuous paths of pixels that run through local minimum energy areas. Removing or inserting pixels along a seam enables users to shrink or enlarge pictures by a wide range, while still retaining all details of the image.

High-density light field reproduction using overlaid multiple projection images

We propose a method to increase the viewing resolution of an autostereoscopic display without increasing the density of microlenses. Multiple projectors are used for the projection images to be focused and overlaid on a common plane in the air behind the microlens array. The multiple overlaid projection images yield multiple light spots inside the region of each elemental lenslet of the microlens array. This feature provides scalable high resolution images by increasing the number of projectors.

Interactive autostereoscopic display with 60 ray directions

We need to design some parameters to determine the specifications of the IP display. The parameters are lens pitch, pixel density of 2D display, lens focus, lens layout, color filter layout of the display, and so on. We think that the maximal spatial frequency and the number of rays are the most important properties needed to produce realistic autostereoscopic images. Therefore, we must decide which parameters to use to increase the number of rays and the maximal spatial frequency. The lens pitch usually determines the maximal spatial frequency of the IP displays. Whereas, the lens pitch and the pixel density of the 2D display determine the number of rays for IP displays. As there is a trade-off between the lens pitch and the pixel density, we must first decide minimum resolution of the IP display, and then decide the lens pitch and the pixel density of the 2D display.

50.1: Invited Paper: Sample Applications Suitable for Features of Integral Videography

I will discuss the features and suitable application for the Integral Videography (IV) display that we developed comparing with two other types of autostereoscopic displays. I will also describe two sample applications that utilize the natural and intuitive 3D expression of the display.

Multiperspective rendering for anime-like exaggeration of joint models

In this paper, representing view-dependent deformation of the perspective effect with a multi-perspective approach is discussed. Traditional hand-drawn animation often constructs a more flexible and impressive depiction than that of computer graphics. Among deformations in handmade drawing, anime has traditionally used deformed depictions that are used to exaggerate the effect of perspective in order to exaggerate the posture of a figure into stylish and dramatic distortions. However, in terms of 3D geometry, the depiction may contain a few contradictions between each part of the body and the background.

Analysis for reproduced light field of 3D displays

A method has been developed for analyzing the characteristics of light fields reproduced by 3D displays and for determining the maximum depth of images on a 3D display. Use of this method will enable identification of which type of 3D display is best for certain applications and will facilitate the design 3D displays.

Perceptual interaction of optical BCI applications

A non-invasive brain-computer Interface (BCI) offers a direct interaction between the human brain and an external device without any surgical risk. Currently, electroencephalography (EEG) has been the most studied non-invasive interface because of its portability and low set-up cost. However, susceptibility to electromagnetic noise has been a barrier to promoting practical non-invasive BCIs in some cases.

Multiple multiperspective rendering for autostereoscopic displays

Autostereoscopic displays, which do not require special glasses or lighting conditions but use selector screen methods such as parallax barriers, lenticular sheets, and lens arrays, show attractive 3D scenes. However these methods sometimes fail to separate each image because of a lack of display resolution or insufficient lens quality. This disarrangement is more highly visible especially at the viewing areas more distant from the actual display plane, where binocular disparity causes much discrepancy of images. Currently, most graphic designers manage to develop scenes by using heuristic procedures in accordance with the limitations of each display system.