Junichi Odagiri

Pupil detection in the presence of specular reflection

In this work we describe a method of pupil detection for subsequent gaze tracking, when specular reflection is present in the image. Gaze tracking commonly uses the spatial relationship between the pupil and corneal reflection, but is not robust when the user is wearing eyeglasses, since light reflected from the surroundings changes the appearance of the pupil. In this research we propose and evaluate a pupil detection method that can perform robustly even in the presence of such reflection.

Data Compression Technology Dedicated to Distribution and Embedded Systems

In distribution and embedded systems, data compression is often used to reduce the size of flash RAM and transmission data, while a rapid decompression speed enables faster rebooting of the compressed program code. We have developed a new data compression algorithm with a high decompression speed and a good compression rate that is equivalent to zlib, the standard technology in use today. We created a LZSS-based algorithm by optimizing the parsing of data strings. LZSS is known as a high decompression speed algorithm useful for embedded systems, and optimal parsing is well known as a method for improving compression rates [1]. Previously, this combination had not been implemented because statistical code length varies during optimal parsing [1]. Our algorithm overcomes this problem by calculating the probability of the literal or the code ( distance and length ) solving the shortest path problem first. It then constructs a simple code set that enables fast decompression using those probabilities and solves the shortest path problem again. Experiments on the standard evaluation data and wireless sensor network program [2] demonstrated that we can achieve a high compression rate equivalent to zlib and a decompression speed that is twice as fast.