The use of high-definition images is increasing and higher resolution images will continue to be popular in the future. High-definition images expand the visual angle of the image field and enhance the presence of the image while maintaining the number of pixels per unit visual angle. We are also able to set a larger number of pixels, which will allow us to perceive the objects in an image with enhanced reality. Thus, stereo images with higher resolution will also be realized, and technologies of operating images will improve further in the future. Eye movement measurement devices are quite common and are utilized not only in research but also as interface tools for personal computers or tablet terminals. In this paper, we discuss recent eye movement measurement experiments on the 3D perception of stereo images. Our primary focus is occlusion caused by depth discontinuity, which is important for 3D perception when displaying stereo pair images.

View full abstract

Recently, a novel method for signal reconstruction, called compressed sensing or sparse modeling, has been attracting considerable attention in signal processing and machine learning. This method has been extended to optimal control, which is called sparse optimal control or dynamical sparse modeling. This control can stop actuators for a time duration, and the control systems can thus achieve effective energy conservation. In this article, we will discuss how sparsity can be utilized in control systems.

View full abstract

The minimum description length principle (MDL principle) is a data-compression-based methodology for optimal estimation and prediction from data. It gives a unifying strategy for designing machine learning algorithms and plays an important role in knowledge discovery from big data. Conventionally, the MDL principle has been extensively studied under the assumption that the information sources are stationary and are represented as regular probabilistic models. This paper first gives a survey of the fundamental concept of the MDL principle. Then, it introduces recent advances in MDL research for the situation where the information sources are nonstationary, irregular, and nonprobabilistic. It also shows trends in the nonasymptotic analysis of the MDL and refers to applications to data mining.

View full abstract

In LSIs for sensor nodes or medical devices, energy dissipation is minimized by reducing the supply voltage to a value near the threshold voltage (Vt) of MOS transistors or even lower than Vt. Design issues in these LSIs are different from those of LSIs operating at a normal voltage. In addition to the difficulty in achieving stable operation, the circuit performance is significantly affected by process and temperature variations. This paper focuses on the logic and memory circuits used in microcontrollers or SoCs operating at ultra-low voltages, and surveys major issues and design techniques. As new topics, two key enablers for ultra-low-voltage operation are also described: a fully depleted silicon-on-insulator (FD-SOI) device and the standard cell memory (SCM) approach.

View full abstract

Sound field simulation is widely used because of the progress in computer environments and the spread of simple and easy simulation algorithms such as the finite-difference time domain (FDTD) method. In this paper, the theoretical basis of the FDTD method for sound field analysis is first described, then the implementation of the FDTD algorithm on a graphics processing unit is described. For the application of sound field simulation to auralization, sound field rendering is described. Sound field rendering is a technique for auralization from three-dimensional numerical models constructed by a computer, with a similar concept to the graphics rendering technique. An example of sound field rendering in collaboration with a head-mounted display is described.

View full abstract