A method for segmentation-free recognition of motions including letters drawn in the air is proposed. Multiple different kinds of motions sharing time periods and space parts can be recognized. The method is robust to different variations of motion, large occlusions and moving background.
In this patents, a handwriting model is proposed to synthesize various handwritten characters while preserving the writer’s individuality from a limited number of training data, using a statistical approach. The proposed model is verified in single- and multiple-stroke characters, such as Arabic numbers, small English letters, and Japanese Kanji letters.
以前の研究は以下のような共通の問題点(限界点)を持っていた。 A. 個々の文字を手で書く際の多様性を１つのモデルで表現できない。 B. 原本の筆跡の個性を維持しながら、多様な手書き文字を生成する技術については前例がない。
In recent years, frequent leakage of important information is becoming a threat of the society, and it is urgently necessary to provide new technologies to complement existing ones. Our new technology has been proposed for this purpose. It is a new steganography technology, and can improve the security through information hiding.
A technology to create a customizable sensor network based on a behavior scenario given by a user, and a method to acquire desired sensing data actively by asking about the information to the senor network.
The heart rate data and body temperature data are measured automatically during sleep, after removing the apparent outliers through proper preprocessing method, Cosinor analysis and Hidden Markov Model are used to estimate the biphasic property in the menstrual cycle in order to identify the stage of ovulation and menstruation.